Amazon Transparency Report

1. Data acquisition methods (Score: 1)

What methods does the developer use to acquire data used to build the model?

Disclosure:

The Amazon Nova models are trained on licensed data, proprietary data, open source datasets, and publicly available data. Our acqusition methods cover: licensing, crawling, synthetic generation and human creation/ annotation.

References:

Not disclosed

Score justification:

The developer clearly states data acquisition methods.

Indicator notes:

Which of the following data acquisition methods does the developer use: 
 (i) acquiring existing public datasets, (ii) crawling the web, (iii) using data acquired via its existing products and services, (iv) licensing existing data from external parties, (v) having humans create or annotate new data, (vi) using models to generate new data,
 or (vii) other data acquisition methods not captured by the above. For example, if the developer uses reinforcement learning from human feedback to train models using model-generated outputs with human preference annotations, this would satisfy categories (v) and (vi). Alternatively, if the developer post-trains its model using off-the-shelf preference data (for example, the Alpaca dataset), this would satisfy category (i).

Example disclosure:

To build our model, we acquire data by crawling the Internet for publicly available data, licensing data from third-parties, and using models to synthetically generate new data. Humans do not create new data nor do we use data from our other products/services to train our model.
2. Public datasets (Score: 0)

What are the top-5 sources (by volume) of publicly available datasets acquired for building the model?

Disclosure:

Our models were trained on data from a variety of sources, including licensed data, proprietary data, open source datasets, and publicly available data where appropriate. We curated data from over 200 languages, with particular emphasis on Arabic, Dutch, English, French, German, Hebrew, Hindi, Italian, Japanese, Korean, Portuguese, Russian, Simplified Chinese, Spanish, and Turkish

References:

Not disclosed

Score justification:

The developer does not provide information about the specific public datasets they use to train their flagship model.

Indicator notes:

We define a source as the entity or means by which the developer acquires data. We define the top-5 sources as the top-5 sources by data volume.

Example disclosure:

We acquire publicly available data from only two sources: The Pile and CommonCrawl.
3. Crawling (Score: 0)

If data collection involves web-crawling, what is the crawler name and opt-out protocol?

Disclosure:

Amazon does not disclose this information publicly.

References:

Not disclosed

Score justification:

The developer does not provide information about crawling involved in acquiring their training data.

Indicator notes:

We award this point for disclosure of the crawler name and opt-out protocols, including if/how they respect the Robots Exclusion Protocol (robots.txt).

Example disclosure:

Our web crawler is named A and information on the opt-out protocol can be found at this URL: ... The CommonCrawl web crawler is named CCBot and information on the opt-out protocol can be found at this URL: https://commoncrawl.org/faq#:~:text=How%20can%20I%20block%20the,%2Dagent%20string%20is%3A%20CCBot.
4. Usage data used in training (Score: 0)

What are the top-5 sources (by volume) of usage data from the developer's products and services that are used for building the model?

Disclosure:

Amazon does not disclose this information publicly.

References:

Not disclosed

Score justification:

The developer does not provide information about the sources of usage data they use to train their flagship model.

Indicator notes:

We define usage data as data collected from the use of a developer's products or services.

Example disclosure:

We use usage data from only two sources: our deployed chatbot X and our online social media platform Y.
5. Notice of usage data used in training (Score: 0)

For the top-5 sources of usage data, how are users of these products and services made aware that this data is used for building the model?

Disclosure:

For Amazon Nova models used via Amazon Bedrock, Amazon Bedrock doesn't store or log prompts and completions and doesn't use prompts and completions to train any AWS models and doesn't distribute them to third parties. For Amazon Nova models used via nova.amazon,com, Amazon records, processes, and retains your Amazon Nova Interactions in the cloud to provide, develop, and improve our services, including artificial intelligence models and to enforce our terms.

References:

Not disclosed

Score justification:

The developer does not list the sources of usage data to clarify if it is only nova.amazon.com, nor if data is specifically used for training models (as opposed to other more generic forms of model improvement).

Indicator notes:

We define usage data notice as the proactive disclosure to users of how their data is used for model development. For example, via a pop-up with a description, a link to the privacy policy, or link to a description of company practices.

Example disclosure:

We notify users of our chatbot X that chatbot interactions are used to train our AI via a pop-up as shown at this URL: ... We notify users of our platform Y about whether their data is used to train our AI via a link to our privacy policy when they sign up for an account.
6. Licensed data sources (Score: 0)

What are the top-5 sources (by volume) of licensed data acquired for building the model?

Disclosure:

The Amazon Nova models were trained on data from a variety of sources, including licensed data, proprietary data, open source datasets, and publicly available data where appropriate. Amazon has announced a licensing agreement with the New York Times that will allow them to use content owned by the NYT for AI-related uses including the training of Amazon's proprietary foundation models. 

References:

Not disclosed

Score justification:

The developer lists that they license data from the New York Times, but does not clarify if this is the sole source of licensed data.

Indicator notes:

We define a source as the entity from which the developer acquires data. For example, the Associated Press is reportedly a source of licensed data for OpenAI.

Example disclosure:

We license data from only three sources: A, B, and C.
7. Licensed data compensation (Score: 0)

For each of the top-5 sources of licensed data, are details related to compensation disclosed?

Disclosure:

Amazon does not disclose this information publicly.

References:

Not disclosed

Score justification:

The developer does not provide information about compensation for licensed data.

Indicator notes:

We award this point if the model developer describes the compensation structure specified in the contract with the data source or indicates they are prohibited from sharing this information if contractually mandated.

Example disclosure:

We compensate A by ... We cannot disclose information on compensation for our relationships with B and C due to contractual terms that prohibit public disclosure.
8. New human-generated data sources (Score: 0)

What are the top-5 sources (by volume) of new human-generated data for building the model?

Disclosure:

We use human generated data as part of three feedback methods to improve our models: 1. Reward modeling: We train models to understand human preferences using techniques like Direct Preference Optimization (DPO) and Proximal Policy Optimization (PPO). 2. Reinforcement Learning from Human Feedback (RLHF): We collect specific human preference data to guide how models learn which outputs are most helpful and appropriate. 3. Red teaming: Our subject matter experts in various fields actively test our models, identifying potential issues and vulnerabilities.

References:

Not disclosed

Score justification:

The developer describes the function of new human-generated data but not its sources.

Indicator notes:

We define a source as the entity or means by which the developer acquires data. For example, Scale AI could be a source of new human-generated data. By new, we mean the data is specifically acquired for the purposes of building the model.

Example disclosure:

We acquire new human-generated data from only two sources: our internal data annotation team and an external vendor, A.
9. Instructions for data generation (Score: 0)

For each of the top-5 sources of human-generated data, what instructions does the developer provide for data generation?

Disclosure:

Amazon does not disclose this information publicly. We do state publicly that we create guidelines that are used to train annotators.

References:

Not disclosed

Score justification:

The developer does not provide instructions they provide to sources of new human-generated data.

Indicator notes:

The instructions should be those provided to the data source. For example, if a third-party vendor works directly with the data laborers to produce the data, the instructions from the developer to this vendor should be disclosed.

Example disclosure:

We instruct our internal data annotation team as follows: ... We instruct vendor A as follows: ...
10. Data laborer practices (Score: 0)

For the top-5 sources of human-generated data, how are laborers compensated, where are they located, and what labor protections are in place?

Disclosure:

Amazon does not disclose this information publicly

References:

Not disclosed

Score justification:

The developer does not provide information about data laborer practices for new human-generated data.

Indicator notes:

For each data source, we require (i) the compensation in either USD or the local currency, (ii) any countries where at least 25% of the laborers are located, and (iii) a description of any labor protections. We will award this point if the developer discloses that it is not aware of data laborer practices.

Example disclosure:

Our internal data annotation team is located in the US, is compensated at 20 USD per hour, and deals with data that does not require specific protections. Our sole external data vendor contracts laborers in Kenya, compensates them at KES 15000 per month, and implements protections for dealing with toxic or unsafe content such as A and B.
11. Synthetic data sources (Score: 0)

What are the top-5 sources (by volume) of synthetic data acquired for building the model?

Disclosure:

Amazon generates synthetic data using propriety models as part of its Automated Red-Teaming efforts to improve robustness of the Nova models. 

References:

Not disclosed

Score justification:

The developer does not describe the sources of synthetic data for model training, namely the specific models used.

Indicator notes:

We define a source of synthetic data as a non-human mechanism (e.g. a machine learning model) used to generate the data.

Example disclosure:

We synthetically generate data using only our previous model X and an early checkpoint of our current flagship model Y.
12. Synthetic data purpose (Score: 1)

For the top-5 sources of synthetically generated data, what is the primary purpose for data generation?

Disclosure:

Amazon generates synthetic data using propriety models as part of its Automated Red-Teaming efforts to improve robustness of the Nova models. Using existing adversial prompts as seeds, Amazon generates additional prompts. We also automatically generate multiturn, multilingual, and multimodal attacks against our system using our Automated Red-Teaming system.

References:

Not disclosed

Score justification:

The developer describes the purpose for synthetic data creation.

Indicator notes:

We define a source of synthetic data as a non-human mechanism (e.g. a machine learning model) used to generate the data.

Example disclosure:

We use model X to generate instruction-tuning data and we use model Y to generate candidate responses that humans select between to provide human preference data for reinforcement learning with human feedback.
13. Data processing methods (Score: 1)

What are the methods the developer uses to process acquired data to determine the data directly used in building the model?

Disclosure:

We used AWS EMR and AWS Batch for data filtering, deduplication, and enrichment pipelines.

References:

Not disclosed

Score justification:

The developer lists three high-level data processing methods.

Indicator notes:

We will award this point for disclosure of all of the methods used to process acquired data. Data processing refers to any method that substantively changes the content of the data. For example, compression or changing the data file format is generally not in the scope of this indicator.

Example disclosure:

We process data in the following six-step pipeline: (i) removal of HTML artifacts, (ii) deduplication, (iii) language identification to retain English data, (iv) removal of CSAM imagery, (v) removal of train-test overlap, and (vi) tokenization.
14. Data processing purpose (Score: 0)

For each data processing method, what is its primary purpose?

Disclosure:

We used AWS EMR and AWS Batch for data filtering, deduplication, and enrichment. An example of data filetering is de-identifying or removing certain types of personal data from our training data, when feasible. Enrichment refers to processes such as the addition or modification of metadata to training data.

References:

Not disclosed

Score justification:

The developer does not describe the purpose for specific data processing methods, but the purpose of data filtering is only partially specified in relation to personal information and the substantive purpose of enrichment is not specified.

Indicator notes:

Data processing refers to any method that substantively changes the content of the data. For example, compression or changing the data file format is generally not in the scope of this indicator.

Example disclosure:

Examples of primary purposes for a data processing method could include: (i) removes low quality data, (ii) removes potentially personal/copyrighted data, (iii) removes product-irrelevant data, (iv) removes toxic data, (v) improves evaluation integrity, or (vi) prepares the data for training the model.
15. Data processing techniques (Score: 0)

For each data processing method, how does the developer implement the method?

Disclosure:

We used AWS EMR and AWS Batch for data filtering, deduplication, and enrichment pipelines.

References:

Not disclosed

Score justification:

The developer does not describe the techniques that implement the data processing steps in adequate depth.

Indicator notes:

Data processing refers to any method that substantively changes the content of the data. For example, compression or changing the data file format is generally not in the scope of this indicator.

Example disclosure:

Examples of how a data processing method is implemented could include: the method (i) is implemented using an in-house regular expression, (ii) is implemented using an in-house tool based on n-gram overlap, (iii) is implemented using a FastText classifier trained on Wikipedia data, (iv) is implemented using hash collisions with the NCMEC database, (v) is implemented by searching for known benchmark canary strings, and (vi) is implemented using tiktoken (https://github.com/openai/tiktoken).
16. Data size (Score: 0)

Is the size of the data used in building the model disclosed?

Disclosure:

We do not disclose token size for model training. However, we provide context length capabilities for each of our understanding models as follows: Nova Pro: 300k tokens Nova Lite: 300k tokens Nova Micro: 128k tokens Nova Premier: 1 million tokens

References:

Not disclosed

Score justification:

The developer does not provide the dataset size, instead providing the context length.

Indicator notes:

To receive this point, the developer should report data size in appropriate units (e.g. bytes, words, tokens, images, frames) and broken down by modality. Data size should be reported to a precision of one significant figure (e.g. 4 trillion tokens, 200 thousand images). The size should reflect data directly used in building the model (i.e. training data) and not data that was acquired but unused, or data used to evaluate the model.

Example disclosure:

We used 3 x 10^12 tokens of text, 1 x 10^6 images, and 5 x 10^5 hours of audio for training.
17. Data language composition (Score: 0)

For all text data used in building the model, what is the composition of languages?

Disclosure:

Our models were trained by curating data from over 200 languages, with particular emphasis on Arabic, Dutch, English, French, German, Hebrew, Hindi, Italian, Japanese, Korean, Portuguese, Russian, Simplified Chinese, Spanish, and Turkish.

References:

Not disclosed

Score justification:

The developer provides a high-level description of languages of emphasis, but this is insufficient precision and does not describe the method for language identification.

Indicator notes:

To receive this point, the developer should report (i) all languages which make up at least 1% of the data and their corresponding proportions and (ii) a brief description of how languages are labeled (if a publicly available tool is used, include a link to the tool). Proportions should be reported to a precision of two significant figures and should describe proportions of documents labeled with some langauge. An "Unknown" category may be included to denote documents where the language could not be identified.

Example disclosure:

English 80%, Spanish 5.0%, French 3.0%, Chinese 2.0%, Unknown 10%. We used a FastText-based classifier trained on Wikipedia data to identify languages.
18. Data domain composition (Score: 0)

For all the data used in building the model, what is the composition of domains covered in the data?

Disclosure:

We do not disclose this information publicly.

References:

Not disclosed

Score justification:

The developer does not provide information about the domain-level composition of the training data.

Indicator notes:

To receive this point, the developer should report the composition of the main domains included in the data used to train the model. This data should be at a level of granularity lower than broad claims about training on "internet data". For example, this could include the proportion of data from e-commerce, social media, news, code, etc. based on the URLs from which the data is sourced. Proportions should be reported to a precision of one significant figure.

Example disclosure:

Social media 40%, code repositories 30%, news articles 20%, e-commerce product listings 5%, scientific papers 5%.
19. External data access (Score: 0)

Does a third-party have direct access to the data used to build the model?

Disclosure:

No, third parties do not have direct access to the data used to build the model. 

References:

Not disclosed

Score justification:

The developer does not provide access to the training data to any external entity, which is insufficient for this indicator because "we will award this point if at least one such entity is named as having direct access to the data".

Indicator notes:

By a third-party, we mean entities that are financially independent of the developer. We will award this point if at least one such entity is named as having direct access to the data. With that said, we may award this point if the developer provides justifications for prohibiting access to narrowly-scoped parts of the data.

Example disclosure:

Third-parties that have direct access to the data include organizations A and B.
20. Data replicability (Score: 0)

Is the data used to build the model described in enough detail to be externally replicable?

Disclosure:

We do not disclose our training dataset list.

References:

Not disclosed

Score justification:

The developer does not provide sufficient information to meaningfully replicate the training data, because of the use of several other types of data beyond publicly available datasets.

Indicator notes:

We will award this point if the description contains (i) a list of all publicly available training data and where to obtain it and (ii) a list of all training data obtainable from third parties and where to obtain it. These conditions refer to criteria 2 and 3 under the OSI Open Source AI v1.0 definition.

Example disclosure:

The listing of publicly available training data can be found at this URL ... and the listing of all training data obtainable from third parties can be found at this URL ...
21. Compute usage for final training run (Score: 0)

Is the amount of compute used in the model's final training run disclosed?

Disclosure:

We do not disclose compute values publicly.

References:

Not disclosed

Score justification:

The developer does not provide information about training compute.

Indicator notes:

Compute should be reported in appropriate units, which most often will be floating point operations (FLOPs), along with a description of the measurement methodology, which may involve estimation. Compute should be reported to a precision of one significant figure (e.g. 5 x 10^25 FLOPs). This number should represent the compute used to train the final model across all model stages.

Example disclosure:

Our model was trained using 5 x 10^25 FLOPs, measured according to the Frontier Model Forum guidance provided at this URL: https://www.frontiermodelforum.org/updates/issue-brief-measuring-training-compute/
22. Compute usage including R&D (Score: 0)

Is the amount of compute used to build the model, including experiments, disclosed?

Disclosure:

Compute used to train Amazon Nova Premier is not disclosed.

References:

Not disclosed

Score justification:

The developer does not provide information about training compute.

Indicator notes:

Compute should be reported in appropriate units, which most often will be floating point operations (FLOPs), along with a description of the measurement methodology, which may involve estimation. Compute should be reported to a precision of one significant figure (e.g. 7 x 10^26 FLOPs). Compared to the previous indicator, this indicator should include an estimation of the total compute used across experiments used towards the final training run for the model (such as including hyperparameter optimization or other experiments), and not just the final training run itself.

Example disclosure:

Our cumulative compute usage involved in building the model was 7 x 10^26 FLOPs, measured according to the Frontier Model Forum guidance provided at this URL: https://www.frontiermodelforum.org/updates/issue-brief-measuring-training-compute/
23. Development duration for final training run (Score: 0)

Is the amount of time required to build the model disclosed?

Disclosure:

Amount of time to train Amazon Nova Premier is not disclosed. 

References:

Not disclosed

Score justification:

The developer does not provide information about training duration.

Indicator notes:

The amount of time should be specified in terms of both the continuous duration of time required and the number of hardware hours used. The continuous duration of time required to build the model should be reported in weeks, days, or hours to a precision of one significant figure (e.g. 3 weeks). The number of hardware hours should be reported to a precision of one significant figure and include the type of hardware hours. No form of decomposition into phases of building the model is required for this indicator, but it should be clear what the duration refers to (e.g. training the model, or training and subsequent evaluation and red teaming).

Example disclosure:

Our model was trained over a period of 90 days using 4x10^4 NVIDIA H100 GPU-days.
24. Compute hardware for final training run (Score: 0)

For the primary hardware used to build the model, is the amount and type of hardware disclosed?

Disclosure:

Although the type and amount of hardware used to train Amazon Nova Premier is not specifically disclosed, Section 6 of the Amazon Nova Technical Report gives details on the hardware infrastructure used to train the Amazon Nova family of models. The Amazon Nova family of models were trained on Amazon’s custom Trainium1 (TRN1) chips, NVIDIA A100 (P4d instances), and H100 (P5 instances) accelerators.

References:

Not disclosed

Score justification:

The developer provides a list of the types of hardware units, but not the count (to the specified precision), which is insufficient for this indicator.

Indicator notes:

In most cases, this indicator will be satisfied by information regarding the number and type of GPUs or TPUs used to train the model. The number of hardware units should be reported to a precision of one significant figure (e.g. 800 NVIDIA H100 GPUs). We will not award this point if (i) the training hardware generally used by the developer is disclosed, but the specific hardware for the given model is not, or (ii) the training hardware is disclosed, but the amount of hardware is not. We will award this point even if information about the interconnects between hardware units is not disclosed.

Example disclosure:

Our model was trained using 1000 NVIDIA H100 GPUs.
25. Compute provider (Score: 1)

Is the compute provider disclosed?

Disclosure:

Based on Section 6 of the Amazon Nova Technical Report, the Amazon Nova family of models is said to be trained on AWS resources including AWS SageMaker, AWS SageMaker-managed Elastic Kubernetes Service (EKS) clusters, AWS File System X (FSx), Amazon Simple Storage Service (S3).

References:

Not disclosed

Score justification:

The developer provides a list of the compute providers within AWS.

Indicator notes:

For example, the compute provider may be the model developer in the case of a self-owned cluster, a cloud provider like Microsoft Azure, Google Cloud Platform, or Amazon Web Services, or a national supercomputer. In the event that compute is provided by multiple sources or is highly decentralized, we will award this point if a developer makes a reasonable effort to describe the distribution of hardware owners.

Example disclosure:

Compute is provided by Google Cloud Platform.
26. Energy usage for final training run (Score: 0)

Is the amount of energy expended in building the model disclosed?

Disclosure:

Energy required to train Amazon Nova Premier is not disclosed.

References:

Not disclosed

Score justification:

The developer does not provide information about the energy and environmental impacts of model training.

Indicator notes:

Energy usage should be reported in appropriate units, which most often will be megawatt-hours (mWh), along with a description of the measurement methodology, which may involve estimation. Energy usage should be reported to a precision of one significant figure (e.g. 500 mWh). No form of decomposition into compute phases is required, but it should be clear whether the reported energy usage is for a single model run or includes additional runs, or hyperparameter tuning, or training other models like reward models, or other steps in the model development process that necessitate energy usage. If the developer is unable to measure or estimate this quantity due to information not being available from another party (e.g. compute provider), we will award this point if the developer explicitly discloses what information it lacks and why it lacks it.

Example disclosure:

Our model was trained using an estimate 1 x 10^4 MWh of energy. To estimate training energy consumption, we multiplied training FLOPs (5 x 10^25) by a conversion factor using NVIDIA A100 GPU information (3.74 × 10^21 FLOPs/MWh) given we train using FP16 with sparsity.
27. Carbon emissions for final training run (Score: 0)

Is the amount of carbon emitted in building the model disclosed?

Disclosure:

Carbon emissions from the training of Amazon Nova Premier is not disclosed.

References:

Not disclosed

Score justification:

The developer does not provide information about the energy and environmental impacts of model training.

Indicator notes:

Emissions should be reported in appropriate units, which most often will be tons of carbon dioxide emitted (tCO2), along with a description of the measurement methodology, which may involve estimation. Emissions should be reported to a precision of one significant figure (e.g. 500 tCO2). No form of decomposition into compute phases is required, but it should be clear whether the reported emissions is for a single model run or includes additional runs, or hyperparameter tuning, or training other models like reward models, or other steps in the model development process that generate emissions. If the developer is unable to measure or estimate this quantity due to information not being available from another party (e.g. compute provider), we will award this point if the developer explicitly discloses what information it lack and why it lacks it. Emissions should correspond with the energy used in the previous indicator.

Example disclosure:

Our model yielded an estimate of 5 x 10^3 tCO2. To estimate training carbon emissions, we multiplied training energy usage (1 x 10^4 MWh) by a 2023 estimate for the US data center carbon intensity (0.375 tCO2/MWh) given the data centers used in training operate in the US.
28. Water usage for final training run (Score: 0)

Is the amount of clean water used in building the model disclosed?

Disclosure:

Water usage from the training of Amazon Nova Premier is not disclosed.

References:

Not disclosed

Score justification:

The developer does not provide information about the energy and environmental impacts of model training.

Indicator notes:

Clean water usage should be in appropriate units, which most often will be megaliters, along with a description of the measurement methodology, which may involve estimation. Clean water usage should be reported to a precision of one significant figure (e.g., 5000ML). No form of decomposition into compute phases is required, but it should be clear whether the reported water usage is for a single model run or includes additional runs, or hyperparameter tuning, or training other models like reward models, or other steps in the model development process that necessitates water usage. If the developer is unable to measure or estimate this quantity due to information not being available from another party (e.g. compute provider), we will award this point if the developer explicitly discloses what information it lacks and why it lacks it.

Example disclosure:

Our model yielded an estimate of 20 ML water. To estimate training water usage, we multiplied training energy usage (1 x 10^4 MWh) by a 2021 estimate for the US data center water efficiency (1.8 ML per 1,000 MWh) given the data centers used in training operate in the US.
29. Internal compute allocation (Score: 0)

How is compute allocated across the teams building and working to release the model?

Disclosure:

Internal compute allocations across teams working on the training and development of Amazon Nova family of models are not disclosed. 

References:

Not disclosed

Score justification:

The developer does not provide information about internal compute allocation.

Indicator notes:

To receive a point, the developer should provide the compute allocated to each team involved in training the model. We understand there might be no clear allocation of compute across different teams; in that case, report an estimate of the compute used over the last year. Compute allocation should be reported to at least one significant figure.

Example disclosure:

- Safety — 15% - Pre-training — 60% - Post-training — 15% - Infrastructure and reliability — 5%
30. Model stages (Score: 1)

Are all stages in the model development process disclosed?

Disclosure:

We define four stages in building Nova Premier: (1) unsupervised pretraining using a mixture of multilingual and multimodal data from licensed, proprietary, open source, and public datasets, (2) Supervised Fine-Tuning (SFT) on instruction-demonstration pairs including multimodal ones, (3) reward model (RM) training based on human preference data, and (4) final alignment through Direct Preference Optimization (DPO) and Proximal Policy Optimization (PPO) to ensure the model follows human preferences in both quality and responsibility.

References:

Not disclosed

Score justification:

The developer lists four high-level stages for model training.

Indicator notes:

Stages refer to each identifiable step that constitutes a substantive change to the model during the model building process. We recognize that different developers may use different terminology for these stages, or conceptualize the stages differently. We will award this point if there is a clear and complete description of these stages.

Example disclosure:

We define five stages in building the model: (1) unsupervised pre-training, (2) supervised instruction tuning, (3) RLHF, (4) domain-specific fine-tuning, and (5) final safety alignment.
31. Model objectives (Score: 1)

For all stages that are described, is there a clear description of the associated learning objectives or a clear characterization of the nature of this update to the model?

Disclosure:

During the pre-training stage, the objective is to optimize model for next-token prediction accuracy. During the supervised fine-tuning stage, the objective is to optimize model correctness and helpfulness for downstream tasks (i.e. instruction following) and specific domains (i.e. coding). During the reward model training stage, the objective is to minimize a pairwise cross-entropy loss so the reward model approximates human preference for a given input. During the final model training stage, the objective is to align the model with human preferences in both quality and responsibility.

References:

Not disclosed

Score justification:

The developer discloses the objective associated with each training stage.

Indicator notes:

We recognize that different developers may use different terminology for these stages, or conceptualize the stages differently. We will award this point if there is a clear description of the update to the model related to each stage, whether that is the intent of the stage (e.g. making the model less harmful), a mechanistic characterization (e.g. minimizing a specific loss function), or an empirical assessment (e.g. evaluation results conducted before and after the stage).

Example disclosure:

During unsupervised pre-training, the objective is next-token prediction. During supervised instruction tuning, we optimize for correctness and helpfulness on labeled tasks. RLHF aligns model outputs with human preference judgments. Domain-specific fine-tuning focuses on improving in-domain capabilities using specialized data (e.g., code or legal text). Final safety alignment reduces disallowed or harmful responses.
32. Code access (Score: 0)

Does the developer release code that allows third-parties to train and run the model?

Disclosure:

Amazon Nova Premier on Amazon Bedrock is released with a variety of sample code and default parameters for using Amazon Nova models on Bedrock. 

References:

Not disclosed

Score justification:

The developer does not provide code for model training, though does release other types of supplementary code.

Indicator notes:

The released code does not need to match the code used internally.

Example disclosure:

We release training and inference code under an Apache 2.0 license at https://github.com/..., enabling others to replicate our core pipeline.
33. Organization chart (Score: 0)

How are employees developing and deploying the model organized internally?

Disclosure:

Amazon's organizational chart for teams supporting the development of Amazon Nova family of models is not disclosed.

References:

Not disclosed

Score justification:

The developer does not provide information about the organization chart relevant to model training.

Indicator notes:

To receive a point, the developer should provide both the internal organization chart for the team developing the model as well as the headcounts (or a proportion of headcounts) by the team.

Example disclosure:

The model team comprises of 63 people, organized as follows: - CEO - Managing Director (Safety) — 24 people - Managing Director (Pre-training) — 12 people - Managing Director (Post-training) — 11 people - Managing Director (API) — 6 people - Director (Infrastructure and reliability) — 7 people - Director (PR and marketing) — 4 people - Director (hiring) — 7 people
34. Model cost (Score: 0)

What is the cost of building the model?

Disclosure:

The cost of developing Amazon Nova Premier is not disclosed.

References:

Not disclosed

Score justification:

The developer does not provide information about the cost to build their flagship model.

Indicator notes:

Monetary cost should be reported in appropriate currency (e.g. USD), along with the measurement methodology, which may involve estimation. Cost should be reported to a precision of one significant figure (e.g. 200 million USD).

Example disclosure:

We spent approximately 200 million USD on building the model: 50 million for data acquisition, 10 million for data processing, 20 million for personnel, 80 million for compute for R&D priced at market rates, and 40 million for compute for the final training run priced at market rates.
35. Basic model properties (Score: 0)

Are all basic model properties disclosed?

Disclosure:

Input modality: Text, Image, Documents, Video Output modality: Text Model components: Multimodal foundation model with integrated safety measures and responsible AI practices. It includes components for processing text, images, and videos. Model architecture: Not fully disclosed, but it's described as a multimodal foundation model with a one-million token context window. Language Support: 200+ languages with emphasis on Arabic, Dutch, English, French, German, Hebrew, Hindi, Italian, Japanese, Korean, Portuguese, Russian, Simplified Chinese, Spanish, and Turkish. Maximum output tokens: 10K.  Maximum context window (tokens): 1M (equivalent to 400-page document or 90-minute video) Configurability: Allows system prompt configuration to define persona, model instructions, and response schemas RAG support: Supports RAG implementation through tool use and knowledge base integration.

References:

Not disclosed

Score justification:

The developer discloses the modalities, but does not fully disclose the components, architecture, and size.

Indicator notes:

Basic model properties include: the input modality, output modality, model size, model components, and model architecture. To receive a point, all model properties should be disclosed. Modalities refer to the types or formats of information that the model can accept as input. Examples of input modalities include text, image, audio, video, tables, graphs. Model components refer to distinct and identifiable parts of the model. We recognize that different developers may use different terminology for model components, or conceptualize components differently. Examples include: (i) For a text-to-image model, components could refer to a text encoder and an image encoder, which may have been trained separately. (ii) For a retrieval-augmented model, components could refer to a separate retriever module. Model size should be reported in appropriate units, which generally is the number of model parameters, broken down by named component. Model size should be reported to a precision of one significant figure (e.g. 500 billion parameters for text encoder, 20 billion parameters for image encoder). Model architecture is the overall structure and organization of a foundation model, which includes the way in which any disclosed components are integrated and how data moves through the model during training or inference. We recognize that different developers may use different terminology for model architecture, or conceptualize the architecture differently; a sufficient disclosure includes any clear, though potentially incomplete, description of the model architecture.

Example disclosure:

Input modality: Text Output modality: Text Model components: Decoder-only model trained using self-supervised learning, followed by supervised fine tuning and RLHF that are used to align the language model to follow users' instructions and be helpful, harmless, and honest. Model size: 70B parameters Model architecture: Autoregressive (causal, decoder only) transformer language model with rotary position embeddings and are trained on the next token prediction task.
36. Deeper model properties (Score: 0)

Is a detailed description of the model architecture disclosed?

Disclosure:

Not disclosed

References:

Not disclosed

Score justification:

The developer does not disclose this.

Indicator notes:

To receive a point, the model architecture should be described in enough detail to allow for an external entity to fully implement the model. Publicly available code or a configuration file for a model training library (e.g., GPT-NeoX) would be a sufficiently detailed description.

Example disclosure:

The configuration file for training our model using a public model training library A can be found at [URL].
37. Model dependencies (Score: 0)

Is the model(s) the model is derived from disclosed?

Disclosure:

Amazon Nova models are not disclosed to be derived from other models.

References:

Not disclosed

Score justification:

The developer does not disclose this.

Indicator notes:

We will award this point for a comprehensive disclosure of the model or models on which the foundation model directly depends on or is derived from, as well as the method by which it was derived (e.g., through fine tuning, model merging, or distillation). Additionally, we will award a point if the developer discloses that the model is not dependent on or derived from any model.

Example disclosure:

This model is a fine tune of Camel-70B. We used the methods described in [PAPER URL] for distillation.
38. Benchmarked inference (Score: 0)

Is the compute and time required for model inference disclosed for a clearly-specified task on clearly-specified hardware?

Disclosure:

It takes 0.9 seconds to receive the first token from the model after an API request is sent, i.e., TTFT(Time to First Token)=0.9 seconds.  The speed of producing subsequent output tokens after the first token is 63 tokens/second. These numbers are reported by Artificial Analysis and based on runtime Bedrock latency figures.

References:

Not disclosed

Score justification:

The developer discloses the time required for inference but does not disclose the compute.

Indicator notes:

The duration should be reported in seconds to a precision of one significant figure (e.g. 0.002 seconds). Compute usage for inference should be reported in FLOPs/second to a precision of one significant figure (e.g. 5 x 10^21 FLOPs/second). The hardware in this evaluation need not be the hardware the developer uses for inference. The developer can report this figure over some known or public dataset.

Example disclosure:

It takes 0.002 seconds and 5 x 10^21 FLOPs/second to generate 100,000 tokens as 5,000 sequences of length 20 given inputs of length 40 from [DATASET URL]. The fixed set of hardware is 8 NVIDIA A100s.
39. Researcher credits (Score: 0)

Is a protocol for granting external entities API credits for the model disclosed?

Disclosure:

Not disclosed

References:

Not disclosed

Score justification:

The developer does not disclose this.

Indicator notes:

A model credit access protocol refers to the steps, requirements, and considerations involved in granting credits to external entities. We will award this point if the developer discloses key details of its protocol, including (i) where external entities can request access to credits (e.g. via an access request form); (ii) explicit criteria for selecting external entities; and (iii) its policy on granting a transparent decision on whether access has been granted within a specified, reasonable period of time. Additionally, we will award a point if the developer discloses that it does not grant external entities API credits.

Example disclosure:

We implement a researcher access program: (i) Access can be requested from [URL] (ii) Any researcher at an accredited research institution is eligible to apply. Decisions are made based on the alignment between the applicant's project description and our target research directions (as described here: [URL]). (iii) Decision notifications are sent within three weeks of the application receipt.
40. Specialized access (Score: 1)

Does the developer disclose if it provides specialized access to the model?

Disclosure:

No, Amazon does not provide specialized access to Nova models, including Amazon Nova Premier.

References:

Not disclosed

Score justification:

The developer discloses that no specialized access is provided.

Indicator notes:

Specialized access could include several categories, such as early access, subsidized access, or deeper access (e.g., to model weights or checkpoints, that are not publicly available). We will award this point if the developer discloses (i) if it provides specialized access and (ii) statistics on the number of users granted access across academia, industry, non-profits, and governments, to one significant figure.

Example disclosure:

We provide early access to the model via API to: (1) 250 academics vetted by our program (2) 0 industry affiliates (3) 0 non-profit affiliates (3) 2 government entities with whom we have signed MoUs We provide no other specialized research access.
41. Open weights (Score: 0)

Are the model's weights openly released?

Disclosure:

No, Nova Premier's model weights are not openly released.

References:

Not disclosed

Score justification:

The weights are not openly released.

Indicator notes:

To receive this point, model weights need to be publicly available at no cost. Developers may receive this point even if there are some restrictions on the external entities that are permitted access (e.g. geographic restrictions), insofar as these restrictions are transparent (e.g. via a license or some high-level description of who has been granted access to the foundation model).

Example disclosure:

Model weights are available on HuggingFace by following this link: [URL]
42. Agent Protocols (Score: 1)

Are the agent protocols supported for the model disclosed?

Disclosure:

Amazon Nova Pro, Lite, and Micro models can be used as agents. Amazon recently launched Strands Agents an open source SDK that takes a model-driven approach to building and running AI agents. Strands supports model integrations (including Nova) via Bedrock and works with MCP and will soon with with A2A.

References:

Not disclosed

Score justification:

The developer discloses agent protocols that the model supports.

Indicator notes:

Agent protocols are specifications that define how autonomous agents exchange messages, context, or function calls with other agents, tools, or services (e.g., Anthropic’s Model Context Protocol (MCP) and Google’s Agent‑to‑Agent (A2A) spec). To earn this point, documentation must enumerate each protocol and describe any deviations or proprietary extensions.

Example disclosure:

We support MCP and A2A for agents built using model A
43. Capabilities taxonomy (Score: 1)

Are the specific capabilities or tasks that were optimized for during post-training disclosed?

Disclosure:

We focus on the following capabilities during post-training: (1) Code generation (2) Retrieval-Augmented Generation (RAG) for retrieving information from reliable, up-to-date sources (3) Video understanding and interpreting content of video scenes (4) Document understanding (5) Function calling for producing outputs that elicit correct responses from application APIs (6) Agentic interactions for multiturn interactions on the customer's behalf

References:

Not disclosed

Score justification:

The developer discloses the capabilities optimized for during post-training.

Indicator notes:

Capabilities refer to the specific and distinctive functions that the model can perform. We recognize that different developers may use different terminology for capabilities, or conceptualize capabilities differently. We will award this point for a list of capabilities specifically optimized for in the post-training phase of the model, even if some of the capabilities are not reflected in the final model.

Example disclosure:

We focus on the following capabilities during post-training: (1) Coding ability (2) Retrieval of information and factuality (3) Multilingual language proficiency on non-English languages (4) Tool-use
44. Capabilities evaluation (Score: 1)

Does the developer evaluate the model's capabilities prior to its release and disclose them concurrent with release?

Disclosure:

We evaluate the Amazon Nova models across a variety of capabilities and share the results in the Technical Report and Model Card. For Amazon Nova Premier, we share performance on public benchmarks across text, multi-modal, and angentic capabilities. Precise quanitifications are available in Section 2.1 of the Model Card Updated: Page 3 of the Amazon Nova Premier Technical Report includes results for Nova Pro and Nova Premier on on text, multimodal, and agentic capabilities for a diverse set of capabilities across different benchmarks. https://assets.amazon.science/e5/e6/ccc5378c42dca467d1abe1628ec9/amazon-nova-premier-technical-report-and-model-card.pdf

References:

Not disclosed

Score justification:

Table 1 specifies benchmarks and results for all capabilities in the taxonomy.

Indicator notes:

The evaluations must contain precise quantifications of the model's behavior in relation to the capabilities specified in the capabilities taxonomy. We will award this point for any clear, but potentially incomplete, evaluation of multiple capabilities.

Example disclosure:

We evaluate capabilities using the following benchmarks: (1) Coding: HumanEval (2) Retrieval: HotPotQA (3) Multilingual performance: MMMLU (4) Tool use: UltraTool
45. External reproducibility of capabilities evaluation (Score: 0)

Are code and prompts that allow for an external reproduction of the evaluation of model capabilities disclosed?

Disclosure:

We provide the prompt templates for all benchmarks at: https://huggingface.co/datasets/amazon-agi/Amazon-Nova-1.0-Premier-evals. Updated: Code and prompts to reproduce evaluations can be found at : https://huggingface.co/datasets/amazon-agi/Amazon-Nova-1.0-Premier-evals

References:

Not disclosed

Score justification:

Prompts and methodology are provided in the HF link but not any code. In addition, the HF link does not cover all of the benchmarks in Table 1. Out of the 16 evaluations specified in Table 1, MBXP, OCRBench-v2, SimpleQA, SWE-Bench Verified are missing from the provided HF link (though, for two of these benchmarks, the developer discloses that they use the official prompts). For OCRBench-v2, the developer specifies "We use the official prompt of the benchmark for evaluation and we report accuracy across all tasks". For SimpleQA, the developer specifies "We compute accuracy (correctness) using the prompt shared in the SimpleQA paper, and GPT-4o-2024-11-20 as a judge". For SWEBench Verified, the developer specifies "Specifically, we use a simple internal agentic scaffold on a 500 instance subset of SWE-bench known as SWE-bench Verified, and we report resolved rate".

Indicator notes:

The released code and prompts need not be the same as what is used internally, but should allow the developer's results on all capability evaluations to be reproduced. The released code must be open source, following the OSI definition of open source.

Example disclosure:

The code and prompts to reproduce our evaluations can be found on this GitHub repository link: [URL]
46. Train-test overlap (Score: 0)

Does the developer measure and disclose the overlap between the training set and the dataset used to evaluate model capabilities?

Disclosure:

Not disclosed

References:

Not disclosed

Score justification:

The developer does not disclose this.

Indicator notes:

We will award this point if, with every capability evaluation for which the developer reports results, the developer reports the overlap between the training set of the model and the dataset used for evaluation, as well as the general methodology for computing train-test overlap (e.g. a description of how n-gram matching was used).

Example disclosure:

We compute train-test overlap using n-gram matching using the procedure described here [URL]. We evaluate the train-test overlap for the following benchmarks: (1) Coding: HumanEval (1.6%) (2) Retrieval: HotPotQA (4%) (3) Multilingual performance: MMMLU (3%) (4) Tool use: UltraTool (9%)
47. Risks taxonomy (Score: 1)

Are the risks considered when developing the model disclosed?

Disclosure:

Yes, the service card discusses the following risks that Nova Premier has been evaluated for: (1) Frontier Risks: Chemical, Biological, Radiological, and Nuclear (CBRN) Weapons Proliferation, Offensive Cyber Operations, Automated AI Research and Development (AI R&D) (2)Safety Risks: Harmlessness, CSAM (3)Fairness (4)Robustness (5) Privacy

References:

Not disclosed

Score justification:

The developer discloses the list of risks considered.

Indicator notes:

Risks refer to possible negative consequences or undesirable outcomes that can arise from the model's deployment and usage. These consequences or outcomes may arise from model limitations (functions that the model cannot perform) or issues with the model's trustworthiness (e.g., its lack of robustness, reliability, calibration). We recognize that different developers may use different terminology for risks, or conceptualize risks differently. We will award this point for a complete list of risks considered, even if some of the risks are not reflected in the final model.

Example disclosure:

We consider the following risks when developing and evaluating the model: (1) Misinformation (2) Harassment (3) Cybersecurity risks (4) Bioweapons design (5) Revealing personally-identifiable information
48. Risks evaluation (Score: 0)

Does the developer evaluate the model's risks prior to its release and disclose them concurrent with release?

Disclosure:

The Amazon Nova Premier Service Card states that, on average, Amazon Nova FMs correctly produce safe responses to over 90% of harmful prompts in a proprietary dataset (2.4K samples) designed to elicit model responses harmful content (for example, self-harm, violence, animal abuse). Updated: (1) Frontier Risks are reported on in Nova Premier’s Frontier Model Safety Framework Risk Assessment * CBRN: WMDP-Bio (0.84), WMDP-Chem (0.66), Protocol-QA-MCQ (0.48), BioLP-Bench (0.23) * Offensive Cyber: Proprietary Knowledge Benchmark (~0.81), Proprietary CTF Benchmark (~0.80) * AI R&D: On the RE-Bench tasks, Nova Premier showed foundational skills in parsing task intent, initiating complex workflows, and debugging, but it was unable to drive fully automated research, and its final solutions were judged to be either non-functional or underperforming. (2) Safety Risks are reported on in Nova Premier’s Service Card * Harmlessness: Amazon Nova FMs correctly product safe reponses on >90% of harmful prompts in a proprietary dataset (2.4K samples) (3-5) Quantiative benchmarks are not reported for Fairness, Robustness, or Privacy. https://www.amazon.science/publications/evaluating-the-critical-risks-of-amazons-nova-premier-under-the-frontier-model-safety-framework

References:

Not disclosed

Score justification:

The disclosed evaluation results do not cover a majority of risks in the taxonomy: categories 1 and 2 are covered in the taxonomy but not 3-5.

Indicator notes:

The evaluations must contain precise quantifications of the model's behavior in relation to the risks specified in the risk taxonomy. We will award this point for clear evaluations of the majority of the states risks.

Example disclosure:

We evaluate the risks for each of the above harms using HarmBench. The results (in terms of mean attack success rate) are: (1) Misinformation: 0.02 (2) Harassment: 0.01 (3) Cybersecurity: 0.10 (4) Bioweapons design (subset of Chemical Biological in HarmBench): 0.12 (5) Revealing personally-identifiable information (subset of General Harm in HarmBench): 0.02
49. External reproducibility of risks evaluation (Score: 0)

Are code and prompts to allow for an external reproduction of the evaluation of model risks disclosed?

Disclosure:

Amazon does not release code / prompts needed to externally produce risk evaluations

References:

Not disclosed

Score justification:

The developer does not disclose this.

Indicator notes:

The released code and prompts need not be the same as what is used internally, but should allow the developer's results on all risk evaluations to be reproduced. The released code must be open-source, following the OSI definition of open-source.

Example disclosure:

The code and prompts to reproduce our evaluations can be found on this GitHub repository link: [URL]
50. Pre-deployment risk evaluation (Score: 0)

Are the external entities have evaluated the model pre-deployment disclosed?

Disclosure:

Yes, the external entities that evaluated the model pre-deployment are explicitly disclosed in the text. Here are the pre-release external evaluators: Red Teaming Firms: ActiveFence Innodata CBRN Capabilities Assessment: Gomes Group (for chemical synonym vulnerabilities) Nemesys (for nuclear facility threats) Deloitte (for biological weapons-related scientific knowledge) Framework Evaluations: Nemesys Insights METR Runtime Performance: Artificial Analysis (described as "an independent entity that benchmarks AI models and hosting providers") Updated: We contract and pay external evaluators for their services in evaluating and assessing our models. These entities include red-teaming firms (ActiveFence, Gomes Group, Innodata), Frontier Capability Assessors (Gomes Group, Nemeysis, Deloitte, METR).

References:

Not disclosed

Score justification:

The developer discloses the external entities that have evaluated the model pre-deployment. The developer also discloses the financial transactions between the parties. However, the developer does not disclose the terms of the analysis.

Indicator notes:

By external entities, we mean entities that are significantly or fully independent of the developer. We will award this point if the developer specifies the entity that carried out the pre-deployment analysis, discloses the terms of the analysis (such as conditions for releasing the evaluation results or the developer's control over the final results), as well as any financial transaction between the parties. We will award this point if the developer discloses no external entities have evaluated the model pre-deployment, or discloses only terms of the analysis where it is not bound by NDA while still naming all external entities.

Example disclosure:

We provide the following parties access to our model for pre-deployment capabilities evaluation: METR. METR has control over the release of the evaluation results (including whether or not to release the results and the contents of the results being released), but must provide the evaluation results to us for review before release. There are no financial transactions between us and METR.
51. External risk evaluation (Score: 1)

Are the parties contracted to evaluated model risks disclosed?

Disclosure:

Yes, the external entities that evaluated the model pre-deployment are explicitly disclosed in the text. Here are the pre-release external evaluators: Red Teaming Firms: ActiveFence Innodata CBRN Capabilities Assessment: Gomes Group (for chemical synonym vulnerabilities) Nemesys (for nuclear facility threats) Deloitte (for biological weapons-related scientific knowledge) Framework Evaluations: Nemesys Insights METR Runtime Performance: Artificial Analysis (described as "an independent entity that benchmarks AI models and hosting providers")

References:

Not disclosed

Score justification:

The developer provides a list of entities which evaluate the model pre-deployment. The developer does not explicitly specify that this list of entities consitutes all of the contracted entities, but it can be reasonably interpreted as such, so we choose to award a point here. We recommend that the developer clarify whether this is the case.

Indicator notes:

We will award this point if the developer discloses statistics regarding all contracted parties that are responsible for evaluating risks (not limited to external entities or pre-deployment evaluation). This includes the number of contracted for-profit or non-profit entities, government entities, independent contractors, and researchers contracted by the developer to evaluate risks. We will award this point if the developer discloses it has no such contracts.

Example disclosure:

Contracted parties responsible for evaluating risks: (1) 2 contracting non-profits (2) 5 independent contractors (3) 0 government entities (4) 20 researchers
52. Mitigations taxonomy (Score: 1)

Are the post-training mitigations implemented when developing the model disclosed?

Disclosure:

We implement the following post-training mitigations to address model risks: 1. Supervised Fine Tuning (SFT) and Learning with Human Feedback (LHF) for model alignment across responsible AI dimensions 2. Runtime input and output moderation systems: - Input moderation to detect and block/modify prompts containing malicious, insecure, illegal material, or alignment bypass attempts - Output moderation to ensure content adheres to Responsible AI objectives 3. Training data safeguards including: - Data review process across training stages - De-identification and removal of certain types of personal data 4. Fine-tuning safeguards to make models resilient against malicious customer fine-tuning that could undermine initial Responsible AI alignment 5. For RLHF training specifically: - Use of a responsible-AI-specific reward model - Training on internally annotated data across all responsible-AI dimensions 6. Watermarking for image and video generation outputs with C2PA metadata 7. Robust model alignment against adversarial inputs, focusing on preventing: - Sensitive data exfiltration - Unauthorized action execution - Service availability degradation - Malicious content generation

References:

Not disclosed

Score justification:

The developer discloses the post-training mitigations implemented.

Indicator notes:

By post-training mitigations, we refer to interventions implemented by the developer during the post-training phase to reduce the likelihood and/or the severity of the model’s risks. We recognize that different developers may use different terminology for mitigations, or conceptualize mitigations differently. We will award this point for a complete list of mitigations considered, even if some of the mitigations are not reflected in the final model. Alternatively, we will award this point if the developer reports that it does not mitigate risk in this way.

Example disclosure:

We implement supervised fine tuning and reinforcement learning with human feedback to address model risks. We use no other methods to address risks.
53. Mitigations taxonomy mapped to risk taxonomy (Score: 1)

Does the developer disclose how the post-training mitigations map onto the taxonomy of risks?

Disclosure:

We use supervised fine tuning (SFT) and reinforcement learning with human feedback (RLHF) to address: (1) Safety, fairness, controllability, veracity and robustness, and privacy and security dimensions and (2) RLHF specifically uses a responsible-AI-specific reward model trained on internally annotated data. We use runtime input and output moderation systems to prevent: (1) Malware, malicious content, and cyber-crime facilitation; (2) Misinformation that undermines public institutions or endangers health; (3) Disrespect, discrimination, or stereotyping towards groups; (4) Inappropriate content including insults, profanity, obscenity; (5) Bias against demographic groups; and (6) Prompt injection and jailbreaking attempts.

References:

Not disclosed

Score justification:

The post-training mitigations are clearly mapped onto the model risks.

Indicator notes:

We will award this point for a complete mapping of the primary risk that each mitigation is meant to address, even if the mitigation potentially maps on to other risks in the taxonomy. Alternatively, we will award this point if the developer reports that it does not mitigate risk.

Example disclosure:

We use supervised fine tuning for general instruction following. We use RLHF to reduce the model's propensity to output information about cybercrimes, bioweapons, disinformation, content harassing someone, and PII.
54. Mitigations efficacy (Score: 0)

Does the developer evaluate and disclose the impact of post-training mitigations?

Disclosure:

Not disclosed

References:

Not disclosed

Score justification:

The developer does not disclose this information.

Indicator notes:

We will award this point if the developer discloses the results on the risk evaluations before and after the post-training mitigations are applied. Alternatively, we will award this point if the developer reports that it does not mitigate risk in this way.

Example disclosure:

Pre-mitigations (measured through mean attack success rate on HarmBench): (1) Misinformation: 0.80 (2) Harassment: 0.91 (3) Cybersecurity risks: 0.56 (4) Bioweapons design (subset of Chemical Biological in HarmBench): 0.62 (5) Personally-identifiable information (subset of General Harm in HarmBench): 0.52 Post-mitigations (measured through mean attack success rate on HarmBench): (1) Misinformation: 0.02 (2) Harassment: 0.01 (3) Cybersecurity risks: 0.10 (4) Bioweapons design (subset of Chemical Biological in HarmBench): 0.12 (5) Personally-identifiable information (subset of General Harm in HarmBench): 0.02
55. External reproducibility of mitigations evaluation (Score: 0)

Are code and prompts to allow for an external reproduction of the evaluation of post-training mitigations disclosed?

Disclosure:

Not disclosed

References:

Not disclosed

Score justification:

The developer does not disclose this information.

Indicator notes:

The released code and prompts need not be the same as what is used internally, but should allow the developer's results on all mitigations evaluations to be reproduced. The released code must be open-source, following the OSI definition of open-source. Alternatively, we will award this point if the developer reports that it does not mitigate risk.

Example disclosure:

We release the code and prompts for reproducing post-training mitigation evaluations at this GitHub link: [URL]
56. Model theft prevention measures (Score: 1)

Does the developer disclose the security measures used to prevent unauthorized copying (“theft”) or unauthorized public release of the model weights?

Disclosure:

The appendix of Amazon Frontier Model Safety Framework discloses details on security measures used to protect senstiive assets and data across Amazon

References:

Not disclosed

Score justification:

The developer discloses the security measures used to prevent model theft.

Indicator notes:

This indicator assesses the developer's disclosures regarding how it addresses the risk that malicious actors or insiders could exfiltrate or replicate proprietary weights. Security measures could include insider threat analysis and detection, in addition to external threat management. Examples of such measures include encryption at rest, key management, remote attestation, or auditing for suspicious queries. We will award a point if the developer discloses specific steps taken to safeguard the model weights or that none are implemented.

Example disclosure:

We store model weights on encrypted volumes with hardware-based key management. We monitor inference queries for suspicious patterns (like repeated attempts to reconstruct weights token-by-token), and we audit all staff access logs monthly.
57. Release stages (Score: 1)

Are the stages of the model's release disclosed?

Disclosure:

No disclosure.  Updated: For the launch of Nova Premier, we launched a Private Beta to internal and external customers for approximately 4-5 weeks before releasing a publicly available GA candidate. Between Private Beta and GA release, we conducted both quantiative and qualitative evaluations to mitigate risks in the GA candidate released to customers. Beta testing is for evaluation purposes prior to GA launch. Private Betas are not released to testers' end users

References:

Not disclosed

Score justification:

The developer discloses the stages of release for Nova Premier.

Indicator notes:

Release stages include A/B testing, release on a user-facing product, GA release, open-weight release, etc. We recognize that the release of a foundation model falls along a spectrum, with many forms of partial release, and that different developers may conceptualize release differently. We will award a point if the developer provides a clear identification of the stages through which the model was released.

Example disclosure:

We began with an internal alpha test for two weeks, followed by a closed beta with selected enterprise partners for one month, then a public waitlisted preview, and finally a general availability release once thresholds on safety benchmarks were met.
58. Risk thresholds (Score: 1)

Are risk thresholds disclosed?

Disclosure:

Amazon's Frontier Model Safety Framework outlines risk thresholds across three domains which frontier models within Amazon are evaluated for prior to deployment: a) Chemical, Biological, Radiological, and Nuclear (CBRN) Weapons Proliferation b) Offensive Cyber Operations c) Automated AI Research and Development (AI R&D) For each domain, a general description of the threshold is provided: CBRN: "AI at this level will be capable of providing expert-level, interactive instruction that provides material uplift (beyond other publicly available research or tools) that would enable a non-subject matter expert to reliably produce and deploy a CBRN weapon." Offensive Cyber Operations: "AI at this level will be capable of providing material uplift (beyond other publicly available research or tools) that would enable a moderately skilled actor (e.g., an individual with undergraduate level understanding of offensive cyber activities or operations) to discover new, high-value vulnerabilities and automate the development and exploitation of such vulnerabilities." Automated AI R&D: "AI at this level will be capable of replacing human researchers and fully automating the research, development, and deployment of frontier models that will pose severe risk such as accelerating the development of enhanced CBRN weapons and offensive cybersecurity methods." Further, Amazon makes a commitment to  not deploy frontier AI models that exceed specified risk thresholds without appropriate safeguards in place.

References:

Not disclosed

Score justification:

The developer discloses risk thresholds.

Indicator notes:

Risk thresholds determine when a risk level is unacceptably high to a developer (e.g. leading to the decision to not release a model), moderately high (e.g. triggering additional safety screening), or low enough to permit normal usage. We will award this point if the developer discloses explicit risk thresholds that clarify (i) which harmful outcomes are being scored, (ii) how the scores are computed (in general terms, not necessarily disclosing internal algorithms), and (iii) what triggers an action to block, delay, or otherwise modify a model's release. Alternatively, we will award a point if the developer discloses that it does not consider explicit risk thresholds during model release.

Example disclosure:

Our risk threshold for biorisks is the ability to autonomously create bioweapons. Current models score a medium: they don't autonomously create bioweapons but could help a skilled practitioner with access to materials in speeding up creation of bioweapons. Risk thresholds higher than medium would delay the model's release until the risk level drops to medium or below.
59. Versioning protocol (Score: 0)

Is there a disclosed protocol for versioning and deprecation of the model?

Disclosure:

Amazon does not publicly disclose versioning protocols for Amazon Nova family of models, however, Amazon Bedrock assigns each model available on Amazon Bedrock a model lifecycle stage.

References:

Not disclosed

Score justification:

The developer discloses information about their deprecation protocol, but does not disclose information about versioning.

Indicator notes:

We will award a point if the developer discloses how model versions are labeled, updated, deprecated, and communicated to users.

Example disclosure:

We version models based on the date of release: e.g., ModelName-11-01-2024. We additionally provide ModelName-latest, corresponding to the latest release. We deprecate versions of models when we plan to remove access to with a six months notice to users. Users should respond to model deprecation by switching to the newest version of the models or an equivalent non-deprecated model. Users can switch to a different model by replacing the model identifier (to e.g., ModelName-latest for the latest version) in API calls or through the Python SDK.
60. Change log (Score: 1)

Is there a disclosed change log for the model?

Disclosure:

Amazon Nova Reel is on its 1.1 version; updates have been published in the accompanying announcement blog. New versions of other Amazon Nova models have not yet been released.  FMTI Team - https://docs.aws.amazon.com/nova/latest/userguide/doc-history.html

References:

Not disclosed

Score justification:

The disclosure provided by the developer lists details only for a single update/release, which is not sufficient for a point. However, the documentation releases in their "User Guide for Amazon Nova" does list new features/performance improvements, which we deem to be a sufficient change log.

Indicator notes:

We will award a point if the developer publishes a version-by-version record of new features, fixes, or performance improvements.

Example disclosure:

On 11/1/2024 (version ModelName-11-01-2024), we improved model reasoning in technical domains. This resulted in a 20-point increase on the MATH benchmark (from 62% to 82%). Past change logs can be viewed at [URL]
61. Foundation model roadmap (Score: 1)

Is a forward-looking roadmap for upcoming models, features, or products disclosed?

Disclosure:

At re:Invent 2024, Amazon pre-announced launches of a speech-to-speech model (launched as Amazon Nova Sonic in April 2025), Amazon Nova Premier (launched April 2025) and an any-to-any modality model coming in 2025.

References:

Not disclosed

Score justification:

In the past, the developer has publicly disclosed the launch of new models.

Indicator notes:

A foundation model roadmap is a transparent statement about how the developer intends to evolve or expand its LLM offerings, including upcoming models, major feature releases, or expanded products based on the model, along with approximate timelines or version milestones. It can be high-level (e.g., “new model Q2 2025”), but must exist publicly.

Example disclosure:

We plan to release ModelX2 in Q2 2025, featuring enhanced multilingual capabilities and improved retrieval. We also aim to launch an enterprise-specific product tier for regulated industries by early 2026.
62. Top distribution channels (Score: 1)

Are the top-5 distribution channels for the model disclosed?

Disclosure:

Amazon provides access to the Amazon Nova models through Amazon Bedrock and through amazon.nova.com. No additional distribution channels are disclosed and Amazon does not publicy share ranking of these two distribution channels.

References:

Not disclosed

Score justification:

The developer discloses two distribution channels and specifies that these are the only two.

Indicator notes:

We define distribution channels to be either an API provider (a pathway by which users can query the model with inputs and receive outputs) or a model distributor (a pathway by which model weights are released). We recognize that distribution channels may arise without the knowledge of the model developer. For example, the weights of a model may be released through one distribution channel and then be distributed through other channels. Distribution channels can be ranked by any reasonable metric (e.g., number of queries, number of downloads, number of users, revenue). A description of the metric should be provided. API providers and model distributors may be ranked separately using different metrics as long as the total number of distribution channels equals five (if five distribution channels exist). For example, the developer may choose to disclose the top-3 API providers (ranked by the number of queries) and the top-2 model distributors (ranked by the number of downloads).

Example disclosure:

We provide API access to the model through A, B, and C. We distribute model weights through D and E. We pick the top-3 API providers based on the average number of queries per month and the top-2 model weight providers based on the average number of downloads per month.
63. Quantization (Score: 0)

Is the quantization of the model served to customers in the top-5 distribution channels disclosed?

Disclosure:

Amazon does not disclose specific details regarding the quantization of its Amazon Nova models that are served to customers. 

References:

Not disclosed

Score justification:

The developer does not disclose this information.

Indicator notes:

We will award this point for a disclosure of the model precision in each of the top-5 distribution channels.

Example disclosure:

We serve the model at 16-bit precision on all distribution channels.
64. Terms of use (Score: 1)

Are the terms of use of the model disclosed?

Disclosure:

Amazon Nova models are governed by nova.amazon.com Terms of Use and AWS Terms of Service

References:

Not disclosed

Score justification:

The developer provides terms of service that appear to apply to the bulk of the distribution channels.

Indicator notes:

We define terms of use to include terms of service and model licenses. We will award this point for a pointer to the terms of service or model license. In the event that model's licenses are written more generally, it should be clear which assets they apply to. We recognize that different developers may adopt different business models and therefore have different types of model licenses. Examples of model licenses include responsible AI licenses, open-source licenses, and licenses that allow for commercial use. Terms of service should be disclosed for each of the top-5 distribution channels. However, we will award this point if there are terms-of-service that appear to apply to the bulk of the model’s distribution channels.

Example disclosure:

Our terms of service are published at https://ourcompany.com/model-tos - these terms cover both our API and all distribution channels for model weights.
65. Distribution channels with usage data (Score: 1)

What are the top-5 distribution channels for which the developer has usage data?

Disclosure:

Amazon Nova Premier is available to customers via Amazon Bedrock and nova.amazon.com. Usage data is not available to Nova developers from usage on Amazon Bedrock. Usage data is available to Nova developers from usage on nova.amazon.com. The nova.amazon.com Terms of Use make customers aware of how usage data from engagements with Nova Premier on nova.amazon.com can be used: https://www.amazon.com/gp/help/customer/display.html?&nodeId=Tsn7s47ytlgjRBHozK

References:

Not disclosed

Score justification:

Clear that Amazon has usage data for these distribution channels

Indicator notes:

We define distribution channels to be either an API provider (a pathway by which users can query the model with inputs and receive outputs) or a model distributor (a pathway by which model weights are released). We recognize that distribution channels may arise without the knowledge of the model developer. For example, the weights of a model may be released through one distribution channel and then be distributed through other channels. Distribution channels can be ranked by any reasonable metric (e.g., number of queries, number of downloads, number of users, revenue). A description of the metric should be provided. We define usage data as any form of developer-exclusive data collected from any of a developer's distribution channel. A developer has access to usage data from a distribution channel if it is able to use that data for downstream purposes (e.g., analytics, training etc.). Usage data may be shared outside of the developer, but it is initially collected by the distribution channel and shared to the developer.

Example disclosure:

We have access to usage data through the distribution channels: A, B, and C.
66. Amount of usage (Score: 0)

For each of the top-5 distribution channels, how much usage is there?

Disclosure:

Amazon does not publicly disclose usage statistics for the Amazon Nova models.

References:

Not disclosed

Score justification:

Company acknowledges no disclosure

Indicator notes:

Usage should be reported as the number of queries over the span of a month, reported to the precision of one significant figure (e.g., 50 million queries).

Example disclosure:

Distribution channel A: 50 million queries. Distribution channel B: 10 million queries. Distribution channel C: 10 million queries.
67. Classification of usage data (Score: 0)

Is a representative, anonymized dataset classifying queries into usage categories disclosed?

Disclosure:

Amazon does not publicly disclose any representative or anonymized datasets that classify Amazon Nova model queries into usage categories.

References:

Not disclosed

Score justification:

Company acknowledges no disclosure

Indicator notes:

Developers may either share a fully public dataset or a partially restricted dataset (e.g., under a research license). We will award this point if there is a clear, aggregated or sample dataset that reveals categories of tasks/queries.

Example disclosure:

We provide quarterly releases of an anonymized dataset that classifies user queries into 20 broad job-related categories. Researchers can request access via [URL]. We ensure no PII is included.
68. Data retention and deletion policy (Score: 0)

Is a policy for data retention and deletion disclosed?

Disclosure:

Our data privacy and data security policies can be found in the AWS Data Privacy FAQ and nova.amazon.com Privacy notice Amazon has published a privacy notice specifically addressing Privacy questions related to training of Amazon models (https://aws.amazon.com/bedrock/amazon-models/privacy/).

References:

Not disclosed

Score justification:

It remains unclear how user data is removed from Amazon models once a user requests deletion or limited processing.

Indicator notes:

A data retention and deletion policy is a policy for removing particular data from the training set and/or preventing it from being used if there is a user or external request (e.g., “right to be forgotten”) that also covers internal data governance. This includes whether there is a formal process to delete or retract data from future training runs and how long raw data is retained. It also clarifies how quickly deletions propagate to the model (e.g., “only in subsequent major model releases”).

Example disclosure:

We honor verified user requests to delete personal data from our training corpus by removing it from any subsequent scheduled retraining. Our data retention policy ensures chat logs are purged after 90 days.
69. Geographic statistics (Score: 0)

Across all forms of downstream use, are statistics of model usage across geographies disclosed?

Disclosure:

Amazon does not disclose statistics of model usage across geographies.

References:

Not disclosed

Score justification:

Company acknowledges no disclosure

Indicator notes:

We will award this point if there is a meaningful, though potentially incomplete or vague, disclosure of geographic usage statistics at the country-level.

Example disclosure:

We share anonymized per-country usage metrics in a publicly accessible dashboard, updated monthly, on this link: [link]
70. Internal products and services (Score: 0)

What are the top-5 internal products or services using the model?

Disclosure:

Amazon does not disclose usage of Amazon Nova models by internal services, however, it has publicly stated that Alexa+ is powered by LLMs from Amazon Bedrock including Amazon Nova models

References:

Not disclosed

Score justification:

Company acknowledges no disclosure

Indicator notes:

An internal product or service is a product or service built by the developer. Products or services can be ranked by any reasonable metric (e.g., number of users, queries, revenue). A description of the metric should be provided.

Example disclosure:

The model is used in products A, B, C, D, and E. We choose products based on the number of montly active users.
71. External products and services (Score: 0)

What are the top-5 external products or services using the model?

Disclosure:

Reference customers for Amazon Nova models can be found on our website.

References:

Not disclosed

Score justification:

Amazon does not state whether these reference customers are owners/operators of the top 5 external products and services

Indicator notes:

An external product or service is a product or service built by a party external to the developer. Products or services can be ranked by any reasonable metric (e.g., number of users, queries, revenue). A description of the metric should be provided. We will award a point if the developer discloses that that it does not have access to such metrics about external products or services.

Example disclosure:

The model is used in products A, B, C, D, and E. We choose products based on the number of montly active users.
72. Users of internal products and services (Score: 0)

How many monthly active users are there for each of the top-5 internal products or services using the model?

Disclosure:

Amazon does not disclose the number of active users for internal products or services using the Amazon Nova models.

References:

Not disclosed

Score justification:

Company acknowledges no disclosure

Indicator notes:

An internal product or service is a product or service built by the developer. The number of users refers to users who engaged or interacted with the model through the internal product or service over the last month or averaged over the last X months (this should be specified). Number of users should be specified to one significant figure (e.g. 100,000).

Example disclosure:

Over the last 6 months, the total monthly active users for our top-5 products using model Y are: Product A: 100,000 users Product B: 30,000 users Product C: 10,000 users Product D: 10,000 users Product E: 10,000 users
73. Consumer/enterprise usage (Score: 0)

Across all distribution channels for which the developer has usage data, what portion of usage is consumer versus enterprise?

Disclosure:

Amazon does not disclose portions of usage from consumer versus enterprises for Amazon Nova models.

References:

Not disclosed

Score justification:

Company acknowledges no disclosure

Indicator notes:

Consumer usage refers to usage by individual consumers. Enterprise usage refers to usage by enterprise customers (including government use). Consumer and enterprise usage should be calculated in terms of the number of queries by or the amount of revenue from consumer or enterprise users. Percentages should be specified to two significant digits (e.g., 12% consumer, 88% enterprise).

Example disclosure:

12% of the usage of model A across all distribution channels is from consumers, 88% is from enterprise users. Of this 88%, 6% is from users at governments. Usage is calculated based on number of queries.
74. Enterprise users (Score: 0)

Across all distribution channels for which the developer has usage data, what are the top-5 enterprises that use the model?

Disclosure:

Amazon does not disclose the top customers for the Amazon Nova models.

References:

Not disclosed

Score justification:

Company acknowledges no disclosure

Indicator notes:

Enterprises should be ranked by the number of queries made or the amount of revenue from usage since the model's release. We will also award this point if the developer indicates it does not have access to enterprise usage data.

Example disclosure:

The top-5 enterprises are A, B, C, D, and E. The enterprises are selected based on the number of queries.
75. Government use (Score: 0)

What are the 5 largest government contracts for use of the model?

Disclosure:

Amazon does not disclose customers for the Amazon Nova models beyond those that are identified on our website.

References:

Not disclosed

Score justification:

Company acknowledges no disclosure

Indicator notes:

This includes known government contracts of enterprise or government-specific products and services that use the model. We will award this point if the developer discloses its top five government contracts ranked monetary value, though the developer may omit contracts where it is under NDA regarding the existence of the contract.

Example disclosure:

The five largest government users of our service, along with their use cases, are: 1. County A is utilizing our product for improving access to internal resources 2. National Lab B is using our model to advance bioscientific research. 3. Federal agency C is using our product to deliver faster, more accurate translation services 4. City D is participating in a pilot program found our product helped reduce the time spent on routine tasks 5. Country E is using our product to summarize legal documents in their lower courts.
76. Benefits Assessment (Score: 0)

Is an assessment of the benefits of deploying the model disclosed?

Disclosure:

Amazon claims that Amazon Nova Micro, Amazon Nova Lite, and Amazon Nova Pro are at least 75 percent less expensive than the best performing models in their respective intelligence classes in Amazon Bedrock. Nova Premier can be used as a teacher model for distillation, which means you can transfer its advanced capabilities for a specific use case into smaller, faster, and more efficient models like Nova Pro, Micro, and Lite for production deployments. This enables customers to improve price and performance on their use cases. In this blog (https://aws.amazon.com/blogs/aws/amazon-nova-premier-our-most-capable-model-for-complex-tasks-and-teacher-for-model-distillation/), a distilled Nova Pro had a 20% higher accuracy for API invocations compared to the base model and consistently matched the performance of the teacher, with the speed and cost benefits of Nova Pro.

References:

Not disclosed

Score justification:

Needs to be a real-world assessment of benefits, not just based on features of the model

Indicator notes:

We will award this point for any quantitative assessment of the benefits or potential benefits of deploying the model.

Example disclosure:

We analyze the impact of using the model in education outcomes using a randomized controlled trial in third grade math assignnments, and find that use in the classroom improves standardized test outcomes by 26%. [Link to report.]
77. AI bug bounty (Score: 1)

Does the developer operate a public bug bounty or vulnerability reward program under which the model is in scope?

Disclosure:

Amazon runs a Vulnerability Research Program (VRP) on HackeOne platform at https://hackerone.com/amazonvrp/policy_scopes. External researchers can use this channel, with the asset being 'GenAI Apps under *.amazon.*' to responsibly disclose urgent security vulnerabilities in GenAI foundation models and applications hosted on Bedrock or nova.amazon.com.

References:

Not disclosed

Score justification:

Amazon has a public bug bounty, states its model(s) are in scope, and the bug bounty says what is out of scope and has tiers

Indicator notes:

We will award this point for a publicly documented bug bounty or vulnerability reward program describing (i) in-scope vulnerabilities (e.g., prompt bypasses, data leaks), (ii) out-of-scope items, (iii) submission process, and (iv) reward tiers or recognition if applicable. We will award a point if the developer discloses it has no AI bug bounty that encourages external researchers to report security, privacy, or adversarial vulnerabilities in the model.

Example disclosure:

We run a bug bounty program with HackerOne. We award up to $5,000 for critical vulnerabilities, such as discovering a major exploit that circumvents our content filters or reveals private data. [link to bug bounty]
78. Responsible disclosure policy (Score: 1)

Does the developer clearly define a process by which external parties can disclose model vulnerabilities or flaws?

Disclosure:

Amazon runs a Vulnerability Research Program (VRP) on HackeOne platform at https://hackerone.com/amazonvrp/policy_scopes. External researchers can use this channel, with the asset being 'GenAI Apps under *.amazon.*' to responsibly disclose urgent security vulnerabilities in GenAI foundation models and applications hosted on Bedrock or nova.amazon.com.

References:

Not disclosed

Score justification:

Amazon has a public bug bounty where submission can be made. No disclosures can be made to third parties, per the bug bounty's RDP: "Thank you for joining us in supporting ethical and responsible disclosure. By participating in this program, you agree not to share publicly or privately any details or descriptions of your findings with any third party."

Indicator notes:

We will award this point for a description of the process external parties can use for responsbly disclosing model vulnerabilities and flaws, which should include (i) what mechanism external parties can use to disclose vulnerabilities or flaws (e.g., a form, an email) and (ii) what process follows a disclosure (e.g., how much time must parties wait until public release). This is often included with a bug bounty, but can also be standalone. We will award a point if the developer discloses it has no responsible disclosure policy.

Example disclosure:

We maintain a responsible disclosure policy at [URL] that describes how external parties can disclose vulnerabilities and flaws in Model A, including a 45-day disclosure window and an official contact for urgent security vulnerabilities.
79. Safe harbor (Score: 0)

Does the developer disclose its policy for legal action against external evaluators conducting good-faith research?

Disclosure:

Amazon's policy regarding potential legal action against external evaluators conducting good-faith research remains undisclosed.

References:

Not disclosed

Score justification:

Company acknowledges no disclosure

Indicator notes:

We will award this point if the developer discloses whether it has a policy committing it to not pursue legal action against external evaluators conducting good-faith research. This should not be only for software security vulnerabilities, but also AI flaws, and it should be based on researcher conduct standards, not at the sole discretion of the company. We will award this point if the developer provides a clear description of its policy regarding such protections for external researchers, or lack thereof.

Example disclosure:

We do not have a policy for researcher protections for good-faith safety research. OR Our policy ensures no legal action against good‐faith researchers who follow our disclosure guidelines, see: [link]
80. Security incident reporting protocol (Score: 1)

Are major security incidents involving the model disclosed?

Disclosure:

We investigate all reported security vulnerabilities affecting Amazon and AWS services, software, and products including GenAI models. The security bulletin can be found here: https://aws.amazon.com/security/security-bulletins/ We would also issue a CVE and GHSA if it the issue meets the criteria of the Amazon CVE Numbering Authority (CNA) to issue those advisories

References:

Not disclosed

Score justification:

The bottom of the security bulletin page has a link for reporting a vulnerability, and the description and page state how they would report what was found

Indicator notes:

A security incident reporting protocol provides post-deployment transparency about serious incidents or breaches. Security incidents refer to incidents where external security threats affect the model (e.g., data breaches or DDoS attacks on the service). We will award this point if the developer states (i) how to submit a security incident report, (ii) how quickly it will respond, and (iii) when and whether results are disclosed. Every incident need not be reported publicly, but the developer must disclose a policy determining how incidents are reported and disclosed.

Example disclosure:

We publish a public ‘Security Incident Report’ on our website for any confirmed security incident affecting the model within 7 days of a patch being implemented. Users and researchers can report incidents via security@ourcompany.com, and we commit to an initial acknowledgment within 48 hours.
81. Misuse incident reporting protocol (Score: 0)

Are misuse incidents involving the model disclosed?

Disclosure:

Amazon does not disclose a misuse incident reporting protocol, however, within its AUP provides a form for users to report suspected abusive activity. 

References:

Not disclosed

Score justification:

Company acknowledges no disclosure

Indicator notes:

A misuse incident reporting protocol provides post-deployment transparency about incidents of misuse involving the model. As opposed to the previous indicator, this indicator is about actors misusing the model to cause real-world harm, such as misinformation operations or cybersecurity attacks. We will award this point if the developer states (i) how to submit a misuse incident report, (ii) how quickly it will respond, and (iii) when and whether results are disclosed. Every incident need not be reported publicly, but there needs to be a policy governing how incidents are reported.

Example disclosure:

We publish a public ‘Misuse Incident Report’ on our website for any confirmed misuse incident within 7 days of a patch being implemented. Users and researchers can report incidents regarding our flagship foundation model via security@ourcompany.com, and we commit to an initial acknowledgment within 48 hours.
82. Post-deployment coordination with government (Score: 1)

Does the developer coordinate evaluation with government bodies?

Disclosure:

We do not coordinate evaluations with any government entities or AI Safety Institutes. 

References:

Not disclosed

Score justification:

Company discloses no such coordination

Indicator notes:

We will award this point if the developer specifies which government bodies it is coordinating with and for what types of post-deployment evaluations. Government bodies include AI Safety Institutes, national security agencies, national labs, and international governmental enties such as UN agencies or the G7. Evaluation here may also include sharing of the developer's proprietary evaluation results for help with interpretation.

Example disclosure:

We do not coordinate with any government entities or AI Safety Institutes. OR We coordinate with the UK AISI for post-deployment evaluation of cyber, CB, and autonomy-related capabilities.
83. Feedback mechanisms (Score: 0)

Does the developer disclose a way to submit user feedback? If so, is a summary of major categories of feedback disclosed?

Disclosure:

(1) For Amazon Nova models access through Amazon Bedrock, users can submit feedback through through the feedback form in the AWS AI Service Card for Nova Premier or via the feedback feature in Amazon Bedrock console.  (2) For Amazon Nova models accessed through nova.amazon.com, users can submit feedback through the thumbs up/down buttons on responses in interactions on nova.amazon.com

References:

Not disclosed

Score justification:

Amazon discloses feedback mechanisms across distribution channels, no summary disclosed

Indicator notes:

We will award this point if the developer (i) discloses how users can submit feedback (e.g., via a form or a thumbs up/thumbs down for model responses) and (ii) discloses aggregated or categorized feedback data (e.g. a categorization of thumbs up and thumbs down data).

Example disclosure:

Users can submit feedback at this url: [URL] We find that users mainly report issues with API call response times, over-refusals from models, and outdated information in model outputs. A detailed categorization of user reports is available at [URL]
84. Permitted, restricted, and prohibited model behaviors (Score: 1)

Are model behaviors that are permitted, restricted, and prohibited disclosed?

Disclosure:

Prohibited behaviors:: Dangerous activities, self-harm, or use of dangerous substances Use, misuse, or trade of controlled substances, tobacco, or alcohol Physical violence or gore Child abuse or child sexual exploitation Animal abuse or trafficking Misinformation that could undermine public institutions or endanger health Malware or content facilitating cybercrime Discrimination or stereotyping Insults, profanity, obscenity, pornography, hate symbols Full nudity outside of scientific/educational contexts Bias based on demographic characteristics Restricted behaviors: While not explicitly labeled as "restricted," the text implies caution in certain areas: Generating content that could be construed as requesting private information Producing outputs that will be directly surfaced to end users without review Use in workflows producing consequential decisions without human oversight From AWS RAI Policy: Prohibitions. You may not use, or facilitate or allow others to use, the AI/ML Services: for intentional disinformation or deception; to violate the privacy rights of others, including unlawful tracking, monitoring, and identification; to depict a person’s voice or likeness without their consent or other appropriate rights, including unauthorized impersonation and non-consensual sexual imagery; for harm or abuse of a minor, including grooming and child sexual exploitation; to harass, harm, or encourage the harm of individuals or specific groups; to intentionally circumvent safety filters and functionality or prompt models to act in a manner that violates our Policies; to perform a lethal function in a weapon without human authorization or control.

References:

Not disclosed

Score justification:

Amazon provides a helpful breakdown of prohibited and restricted behaviors.

Indicator notes:

We refer to a policy that includes this information as a model behavior policy, or a developer's policy on what the foundation model can and cannot do (e.g. such a policy may prohibit a model from responding to NSFW content). We recognize that different developers may adopt different business models and that some business models may make enforcement of a model behavior policy more or less feasible. We will award this point if at least two of the three categories (i.e. permitted, restricted, and prohibited model behaviors) are disclosed. Alternatively, we will award this point if the developer reports that it does not impose any restrictions on its model's behavior in this way.

Example disclosure:

We allow responses from Model A that include broad Q&A, restrict sexual or harassing content, and prohibit facilitating illegal or violent acts. More details can be found in our guidelines for model behavior here: [link]
85. Model response characteristics (Score: 1)

Are desired model response characteristics disclosed?

Disclosure:

We configure Nova Premier's responses to: - Provide concise answers to simple questions when information is directly available - Include more details for yes/no questions - Use logical reasoning for multi-hop reasoning questions - Be transparent about information gaps by stating when exact answers cannot be found - Include citations to support responses using markers like %[1]%, %[2]%, %[3]% - Avoid completing prompts that could request private information - Adhere to responsible AI objectives through runtime moderation - Maintain user privacy by not storing or sharing customer prompts and completions - Generate responses that align with Amazon's responsible AI dimensions including safety, fairness, controllability, veracity and robustness

References:

Not disclosed

Score justification:

Amazon provides a list of how Nova Premier's responses are configured

Indicator notes:

Model response characteristics include default behaviors or behaviors that the developer steers the model to take. These may include being helpful, taking an objective point of view, or using tools only when necessary. We will award points for a clear description of desired model response characteristics or a statement that there are no such characteristics.

Example disclosure:

We configure responses from Model A to be factual, neutral, and contextually helpful, avoiding personal or biased opinions. More details can be found in our guidelines for model behavior here: [link]
86. System prompt (Score: 0)

Is the default system prompt for at least one distribution channel disclosed?

Disclosure:

No, the system prompt for Amazon Nova Premier is not disclosed for the model accessed through Amazon Bedrock or through nova.amazon.com

References:

Not disclosed

Score justification:

Company acknowledges no disclosure

Indicator notes:

A system prompt is defined as the prompt provided to the system by default that guides the system's behavior. We will award this point for the disclosure of the verbatim text of the full system prompt as well as an explanation for the context in which the system prompt is used.

Example disclosure:

We disclose our default prompt for Model A via our chat interface: ‘You are a helpful AI assistant providing clear, accurate, and policy‐compliant responses.’
87. Intermediate tokens (Score: 1)

Are intermediate tokens used to generate model outputs available to end users?

Disclosure:

Nova Premier makes its intermediate tokens (chain-of-thought reasoning) available to users when instructions are given, and users are advised to use specific instructions to keep the thinking brief and contain it within tags. This is summarized from Amazon Nova's User Guide 

References:

Not disclosed

Score justification:

Amazon discloses COT available with prompting

Indicator notes:

Intermediate tokens are defined as any tokens generated by the model before the final output is shown to the user, such as model chains of thought. We will also award this point if a summary of intermediate tokens is made available to end users. If intermediate tokens or summaries are not made available, the developer should provide a justification.

Example disclosure:

Model A is trained to generate intermediate chain-of-thought reasoning, but we withhold most chain-of-thought tokens from final user-facing responses to prevent model distillation. We do disclose chains-of-thought for a small set of research collaborators under NDA.
88. Internal product and service mitigations (Score: 1)

For internal products or services using the model, are downstream mitigations against adversarial attacks disclosed?

Disclosure:

To help prevent potential misuse, Amazon Bedrock implements automated abuse detection mechanisms. These mechanisms are fully automated, so there is no human review of, or access to, user inputs or model completions. To learn more, see Amazon Bedrock Abuse Detection in the Amazon Bedrock User Guide.

References:

Not disclosed

Score justification:

Amazon automatically scans user inputs

Indicator notes:

An internal product or service is a product or service built by the developer. Adversarial attacks include prompt injection, jailbreaking, or malicious queries. Mitigations against adversarial attacks might include specialized prompt filtering, content scanning, or real-time monitoring of queries or accounts. We will award this point if the developer discloses a clear statement of methods used (e.g., a specialized prompt sanitizer or adversarial pattern detector), or if the developer states it does not implement such product-level mitigations against adversarial attacks.

Example disclosure:

In our chatbot, we implement a second-stage content filter that checks user inputs for disallowed topics and attempts to sanitize adversarial prompts. We also log suspicious prompts for manual review.
89. External developer mitigations (Score: 1)

Does the developer provide built-in or recommended mitigations against adversarial attacks for downstream developers?

Disclosure:

Amazon Nova's User Guide  provides insights into downstream security controls and also provides recommendations to implement model content guardrails via System Prompt field. The developer also has an option to use AWS Bedrock Guardrails at the application layer to mitigate prompt injection attacks, and block harmful content or leakage of sensitive information. 

References:

Not disclosed

Score justification:

AWS Bedrock Guardrails suffices

Indicator notes:

Downstream developers are developers who access the model through a distribution channel. Adversarial attacks include prompt injection, jailbreaking, or malicious queries. Mitigations against adversarial attacks that developers might build in or recommend include content filtering endpoints and recommended prompt templates. We will award this point if the developer discloses (i) technical mitigations (e.g., a developer provided moderation API or classifier) it offers or implements, (ii) recommended best practices or libraries for downstream developers, or (iii) an explicit statement that it does not build or recommend any particular downstream mitigations in this way..

Example disclosure:

Our API includes an optional parameter that will automatically filter user prompts and model outputs for hateful or disallowed content. We also publish guidelines for building robust chat interfaces that resist common prompt injections.
90. Enterprise mitigations (Score: 1)

Does the developer disclose additional or specialized mitigations for enterprise users?

Disclosure:

For enterprise users using Nova via Amazon Bedrock, Bedrock's Abuse Detection mitigates potential misuse to uphold Responsible AI.

References:

Not disclosed

Score justification:

Description of mitigations is sufficient

Indicator notes:

Enterprise users are, for example, large organizations with dedicated service agreements or users of enterprise-specific API deployments or products and services. Additional or specialized mitigations may address enterprise needs such as data privacy controls, advanced prompt/response monitoring, or compliance checks with regulations such as GDPR or HIPAA. Additional or specialized mitigations may include single-tenant deployments, custom filters for specific regulated industries, or advanced logging for compliance. We will award a point if the developer at least describes these mitigations or states that it does not provide such additional or specialized enterprise mitigations.

Example disclosure:

Our enterprise offering for Model A includes a dedicated environment with stricter filtering, a HIPAA-compliant data retention policy, and the ability for enterprise admins to define custom blacklisted topics that the model must refuse.
91. Detection of machine-generated content (Score: 1)

Are mechanisms that are used for detecting content generated by this model disclosed?

Disclosure:

Amazon Nova Premier exclusively generates text outputs, which do not carry any watermkaring. Amazon Nova Reel (video generation) and Amazon Nova Canvas (image generation) models support watermark injection (Canvas, Reel) and C2PA (Canvas).

References:

Not disclosed

Score justification:

Amazon discloses no watermark for text outputs and different watermarks for video and image models

Indicator notes:

A mechanism for detecting machine-generated content might include storing a copy of all outputs generated by the model to compare against, implementing a watermark on model outputs, adding cryptographic metadata (such as C2PA), or training a detector post-hoc to identify such content. We will award this point if any such mechanism is disclosed or if the developer reports that it does not have or use any such mechanism.

Example disclosure:

We train a classifier using model generations and human-written text to identify machine-generated content from Model A and our other models.
92. Documentation for responsible use (Score: 1)

Does the developer provide documentation for responsible use by downstream developers?

Disclosure:

For use of Amazon Nova Premier on Amazon Bedrock, Amazon provides documentation for responsible use by developers including AI Service cards, a suite of tools for building tools responsibly such as Amazon Guardrails, SageMaker Clarify, and ML Governance, and the AWS Responsible AI Policy. For use of Amazon Nova Premier on nova.amazon.com, Amazon provides an Acceptable Use Policy and Terms of Use.

References:

Not disclosed

Score justification:

AWS Responsible AI Policy suffices

Indicator notes:

To receive a point, the developer should provide documentation for responsible use. This might include details on how to adjust API settings to promote responsible use, descriptions of how to implement mitigations, or guidelines for responsible use. We will also award this point if the developer states that it does not provide any such documentation. For example, the developer might state that the model is offered as is and downstream developers are accountable for using the model responsibly.

Example disclosure:

Our Developer Documentation Hub consolidates integration guides, responsible‐use guidelines, and best practices: [link]
93. Permitted and prohibited users (Score: 1)

Is a description of who can and cannot use the model on the top-5 distribution channels disclosed?

Disclosure:

For usage of Amazon Nova models on AWS, no explicit permitted/prohibited users are described. For usage of Amazon Nova models on nova.amazon.com, usage requirements are stated in Section 1.3 of the Terms of Use. Permitted users must be 18 years old. 

References:

Not disclosed

Score justification:

Clear description of permitted/prohibited users

Indicator notes:

We will award this point for a description of the company's policies for permitted and prohibitted users on its top-5 distribution channels. We will award this point if the developer has a more general acceptable use policy that it confirms applies across these distribution channels. We will award this point if there are no restrictions on users.

Example disclosure:

We allow usage by individuals 13 years of age or older who accept our Terms of Service. We prohibit use by export controlled entities or persons on denied-parties lists or in countries under U.S. embargo. We also reserve the right to restrict use if users engage in targeted harassment. For example, we only permit users over 13 with valid credentials, and prohibit usage from OFAC-sanctioned regions. We do not allow state-sponsored disinformation agencies to access our services.
94. Permitted, restricted, and prohibited uses (Score: 1)

Which uses are explicitly allowed, conditionally permitted, or strictly disallowed under the acceptable use policy for the top-5 distribution channels?

Disclosure:

For Amazon Nova models on Bedrock, prohibited uses are documented in the AWS Acceptable Use Policy and AWS Responsible AI Policy. For Amazon Nova models on nova.amazon.com, there is a separate Acceptable Use Policy

References:

Not disclosed

Score justification:

Clear disclosure of the AUPs

Indicator notes:

We will award this point for a rough characterization of two or more of permitted, restricted, and prohibited uses across the top-5 distribution channels. We will award this point if the developer has a more general acceptable use policy that it confirms applies across these distribution channels. We will award this point if there are no restrictions on users.

Example disclosure:

Permitted uses include general conversational queries, brainstorming, and coding assistance. Restricted uses include adult or violent content that requires caution or additional review. Prohibited uses include facilitating illicit activity, disinformation campaigns, or harassment. For example, we permit typical user requests like Q&A, text generation, and educational uses. We restrict content that depicts graphic violence or sexual content by applying additional filters. We prohibit any use aiming to conduct unlawful surveillance, promote extremist violence, or defraud others.
95. AUP enforcement process (Score: 1)

What are the methods used by the developer to enforce the acceptable policy?

Disclosure:

The Amazon Nova family of models leverages Bedrock Abuse Detection service (including CSAM detection) to prevent potential misuse by users on Amazon Bedrock and amazon.nova.com

References:

Not disclosed

Score justification:

Some description of Bedrock Abuse Detection

Indicator notes:

We will award this point if the developer discloses the processes (automated or manual) it uses to detect, review, and respond to potential acceptable use policy violations. We will award this point for a reasonable best-effort attempt to provide the bulk of this information, though one line indicating the developer reserves the right to terminate accounts is insufficient. Alternatively, we will award this point if the developer reports that it does not use such methods to enforce its acceptable use policy.

Example disclosure:

We combine automated checks with human review for severe or repeated violations, issuing warnings or suspensions after repeat violations.
96. AUP enforcement frequency (Score: 0)

Are statistics on the developer's AUP enforcement disclosed?

Disclosure:

Amazon does not publish statistics on Acceptable Use Policies for Amazon Bedrock or nova.amazon.com.

References:

Not disclosed

Score justification:

Company acknowledges no disclosure

Indicator notes:

We will award this point if the developer discloses enforcement statistics (e.g., violation counts or actions taken) from its enforcement of its acceptable use policy. Alternatively, we will award this point if the developer reports that it does not enforce its acceptable use policy.

Example disclosure:

We publish a quarterly enforcement report detailing violation counts by prohibited use category and the corresponding actions taken at [LINK]
97. Regional policy variations (Score: 1)

Are differences in the developer's acceptable use or model behavior policy across geographic regions disclosed?

Disclosure:

For use of Amazon Nova models on Bedrock, usage is governed by AWS Customer Agreement and the AWS service Terms. In the AWS Customer Agreement, differences in jurisdictions are mentioned (Section 3, Section 11). For use of Amazon Nova models on nova.amazon.com, the Terms of Use mention restriction of usage to the US (Section 2.4)

References:

Not disclosed

Score justification:

Variations in policy by region are pointed to

Indicator notes:

We will award this point if the developer discloses distinctions in its AUP or MBP and provides examples of differences in multiple specific regions, or states that no differences exist. For example, some jurisdictions impose content restrictions beyond those in the developer’s global policy that may necessesitate local deviations.

Example disclosure:

In the EU, our model automatically omits certain categories of political content to comply with local election laws. In all other regions, we follow the general global AUP at [URL].
98. Oversight mechanism (Score: 1)

Does the developer have an internal or external body that reviews core issues regarding the model prior to deployment?

Disclosure:

In Amazon's Frontier Model Safety Framework, it commits to incorporate of the Safety Framework into the Amazon-wide Responsible AI Governance Program.

References:

Not disclosed

Score justification:

The developer provides a description of its oversight mechanism in its Frontier Model Safety Framework

Indicator notes:

We will award this point if the developer discloses that is has such an internal or external body and provides some description of its scope, or alternatively if the developer discloses that it has no such body. An oversight mechanism covers governance structure beyond mere external risk evaluation, asking whether a formal body regularly reviews design and deployment decisions. Core issues may include model objectives, data usage, or risk mitigation.

Example disclosure:

We convene a monthly advisory board of ethicists, civil society representatives, and academics to review training processes and identify new risks. The board's recommendations regarding deployment are not binding.
99. Whistleblower protection (Score: 0)

Does the developer disclose a whistleblower protection policy?

Disclosure:

Amazon does not have public disclosures about its whistlwblower protection policy. 

References:

Not disclosed

Score justification:

Company acknowledges no disclosure

Indicator notes:

We will award this point if the developer discloses (i) the existence of a whistleblower protection policy, (ii) what protections are afforded to whistleblowers, (iii) how reports are handled and investigated, and (iv) any external oversight of the whistleblower protection process. This might include protections for whistleblowers who report safety, ethical, or legal concerns related to the model. We will also award this point if the developer discloses that it has no such policy.

Example disclosure:

We maintain a whistleblower protection policy that prohibits retaliation against employees who report safety or ethical concerns about our models. Reports can be submitted anonymously through our ethics hotline, are reviewed by an independent board committee, and whistleblowers are entitled to legal representation provided by the company. Our policy is audited annually by an independent ethics consultancy.
100. Government commitments (Score: 0)

What commitments has the developer made to government bodies?

Disclosure:

Amazon has publicly committed to collaboration as part of the G7 AI Hiroshima Process Code of Conduct, and the AI Safety Summits in the U.S. and Seoul

References:

Not disclosed

Score justification:

The developer discloses a number of government commitments it has made, though the list is incomplete as it has also signed onto the White House Voluntary Commitments.

Indicator notes:

We will award this point if the company provides an exhaustive list of commitments it has made to government bodies in the jurisdictions where it offers its model.

Example disclosure:

We have committed to the White House Voluntary Committments and the Seoul Committments.