# RAISP02-BP05 Embed provenance indicators into core AI system outputs
<a name="raisp02-bp05"></a>

 Address release criteria for transparency by building provenance indicators directly into your AI system. Providing machine readable labels for audio and imagery outputs is one of the approaches. 

 **Level of risk exposed if this best practice is not established:** High 

## Implementation considerations
<a name="implementation-considerations-68"></a>

1.  Consider the necessity and utility of embedding machine-readable labels into AI-generated content such as images, audio, and video that clearly identifies the content as AI-generated. 

1.  Consider whether to create provenance chains that track the details such as name of the AI system provider, name of the AI system, time stamp of synthetic output generation, and unique identifiers  so that users can trace how content was created and modified. The level of detail provided should balance verification value against data costs, security impacts, and risks of disclosing proprietary system details. 

1.  If your system outputs machine readable labels , provide capabilities that let users check that content has originated from your system. For example, Amazon Bedrock provides customers with the capability to check if an image was generated by Amazon Nova Canvas or Amazon Titan Image Generator via a publicly available tool, [Content Credentials Verify](https://contentcredentials.org/verify). 

## Resources
<a name="resources-65"></a>

 **Related documents:** 
+  [Considerations for addressing the core dimensions of responsible AI for Amazon Bedrock applications](https://aws.amazon.com/blogs/machine-learning/considerations-for-addressing-the-core-dimensions-of-responsible-ai-for-amazon-bedrock-applications/) 
+  [Evaluate models or RAG systems using Amazon Bedrock Evaluations – Now generally available](https://aws.amazon.com/blogs/machine-learning/evaluate-models-or-rag-systems-using-amazon-bedrock-evaluations-now-generally-available/) 
+  [Amazon Titan Image Generator and watermark detection API are now available in Amazon Bedrock](https://aws.amazon.com/blogs/aws/amazon-titan-image-generator-and-watermark-detection-api-are-now-available-in-amazon-bedrock/) 
+  [ISO/IEC 42001:2023 A.8.2 System documentation and information for users](https://www.iso.org/standard/42001) 
+  [Thorn and All Tech Is Human Forge Generative AI Principles with AI Leaders to Enact Strong Child Safety Commitments](https://www.thorn.org/blog/generative-ai-principles/) 

 **Related videos:** 
+  [Amazon Titan Image Generator Demo - Watermark Detection \| Amazon Web Services](https://www.youtube.com/watch?v=M5Vqb3UoXtc) 

 **Related tools:** 
+  [Watermark detection for Amazon Titan Image Generator now available in Amazon Bedrock](https://aws.amazon.com/about-aws/whats-new/2024/04/watermark-detection-amazon-titan-image-generator-bedrock/) 
+  [Generating images with Amazon Nova Canvas](https://docs.aws.amazon.com/nova/latest/userguide/image-generation.html) 
+  [Amazon Nova – AWS AI Service Card](https://docs.aws.amazon.com/ai/responsible-ai/nova-canvas/overview.html)