AWS wellarchitected medium security documentation change
Summary
Added generative AI monitoring considerations including model behavior tracking, response validation, and AI capability misuse detection. Enhanced guidance for monitoring AI endpoint configurations and data access patterns.
Security assessment
The change explicitly addresses security monitoring for emerging AI threats by adding guardrails for response validation and tracking data access patterns. Specific evidence includes additions like 'potential misuse of AI capabilities' and 'implementing guardrails for response validation', which directly relate to mitigating security risks in AI systems.
Diff
diff --git a/wellarchitected/latest/financial-services-industry-lens/fsisec05.md b/wellarchitected/latest/financial-services-industry-lens/fsisec05.md index b1eb70eaa..e8d573a03 100644 --- a//wellarchitected/latest/financial-services-industry-lens/fsisec05.md +++ b//wellarchitected/latest/financial-services-industry-lens/fsisec05.md @@ -9 +9 @@ FSISEC05-BP01 Track configuration changesFSISEC05-BP02 Detect unusual and unauth -Financial services organizations require in-depth visibility into the security of their infrastructure and applications. Achieving this high level of visibility requires the collection of logs and audit trails and the reservation of these logs for analytics and reporting. AWS services and partners' cloud-native solutions help you implement real-time monitoring in your environment for security threats and alerting on threats once detected. +Financial services organizations require in-depth visibility into the security of their infrastructure and applications. Achieving this high level of visibility requires the collection of logs and audit trails and the reservation of these logs for analytics and reporting. AWS services and partners' cloud-based solutions help you implement real-time monitoring in your environment for security threats and alerting on threats once detected. With generative AI systems, monitoring extends to model behaviors, response validation, and potential misuse of AI capabilities. @@ -15 +15,3 @@ As part of monitoring the environment against threats, it is critical to identif -Use [AWS Config](https://docs.aws.amazon.com/config/latest/developerguide/WhatIsConfig.html) to audit and evaluate the configuration settings of your AWS resources. AWS Config continually tracks the configuration changes that occur in your resources, and by using [AWS Config](https://docs.aws.amazon.com/config/latest/developerguide/evaluate-config_use-managed-rules.html) [Managed Rules](https://docs.aws.amazon.com/config/latest/developerguide/evaluate-config_use-managed-rules.html), it checks to see if these changes comply with the your defined desired state. This allows you to identify and correct configuration deviations as soon as they happen, and also helps the second and third lines of defense respond quickly. +Use [AWS Config](https://docs.aws.amazon.com/config/latest/developerguide/WhatIsConfig.html) to audit and evaluate the configuration settings of your AWS resources. AWS Config continually tracks the configuration changes that occur in your resources, and by using [AWS Config Managed Rules](https://docs.aws.amazon.com/config/latest/developerguide/evaluate-config_use-managed-rules.html), it checks to see if these changes comply with the your defined desired state. This allows you to identify and correct configuration deviations as soon as they happen, and also helps the second and third lines of defense respond quickly. + +For generative AI systems, establish comprehensive monitoring of model endpoint configurations, prompt catalog changes, and AI service policy modifications while implementing guardrails for response validation and tracking data access patterns across AI workflows.