AI Security Maturity Model·AWS v1.0.0

Drag to explore · pinch to zoom
Home / Matrix / Capability 6 / c6.7
Enhanced Logging and MonitoringDefense-in-depth layer: AI Application Capability 6 · AI applicationsUse case: AI that answers 👤 CustomerAWS Shared Responsibility Model: who implements and operates this control. Customer = you own it; AWS = AWS operates it and you inherit it; Shared = both have obligations.

c6.7 Monitor AI application usage with attributed audit logs

Log AI application access and usage with user attribution and monitor for anomalous behavior, so misuse, compromised credentials, and policy violations are detectable.

Foundational
Enhanced
Advanced
Effort: Medium Impact: Medium GenAI scopesFrom the AWS Generative AI Security Scoping Matrix.: 3 · Pre-trained models

Implementation guidance

How to implement on AWS

Enable CloudTrail (organization trail) to log AI app access and usage with user-context attribution. Monitor usage patterns, error rates, and adoption with CloudWatch metrics and alarms, and use app-specific analytics (Q Developer usage analytics, Quick audit logging, Kiro telemetry). Apply per-user rate limits and usage quotas.

Acceptance criteria

AI app access is logged with user attribution and centralized; usage anomalies raise CloudWatch alarms; per-user rate limits/quotas are enforced.

Practical implementation

How this control is implemented in each reference architecture:

AI-app access is logged with user attribution via the org CloudTrail to the Log archive, usage anomalies alarm in CloudWatch, and per-user rate limits are enforced.

  1. Enable the organization CloudTrail to log AI-app access and usage with user-context attribution, forwarded to the Log archive.
    Application account → Log archive CloudTrail Amazon S3
  2. Monitor usage patterns, error rates, and adoption with CloudWatch metrics/alarms, complemented by app-specific analytics (Q Developer usage analytics, Quick audit logging, Kiro telemetry).
    Application account CloudWatch
  3. Apply per-user rate limits and usage quotas.
    Application account AWS WAF API Gateway

How to validate: AI-app access is logged with user attribution and centralized in the Log archive; usage anomalies raise CloudWatch alarms; per-user rate limits/quotas are enforced.

Avoid: Logging app usage without tying it to a user. Attribution is what lets you investigate a compromised account or misuse.

Generative AI OU (multi-account) →

In-account CloudTrail with user attribution, CloudWatch anomaly alarms, and per-user rate limits make AI-app usage auditable.

  1. Log AI-app access and usage with user-context attribution via CloudTrail into the in-account Object-Lock bucket.
    In-account logging CloudTrail Amazon S3
  2. Monitor usage, error rates, and adoption with CloudWatch and app-specific analytics.
    In-account logging CloudWatch
  3. Apply per-user rate limits and quotas.
    Application tier (IAM role) AWS WAF API Gateway

How to validate: AI-app access is logged with user attribution; usage anomalies raise alarms; per-user rate limits/quotas are enforced.

Avoid: No usage analytics or rate limits. You cannot spot abuse or a compromised credential without attributed usage data.

Single account (entry point) →

Threats addressed

  • Compromised credentials or anomalous usage going undetected.
  • No attributed audit trail of AI application access for investigations.

AWS services

AWS CloudTrail Amazon CloudWatch

Framework mappings

OWASP GenAI/LLM Top 10 (2026)

  • LLM02 Sensitive Information Disclosure
  • LLM06 Unbounded Consumption

NIST AI Risk Management Framework (AI 100-1)

  • MEASURE.3 Mechanisms for tracking AI risks are in place
  • MANAGE.4 Risk treatments, monitoring and response are documented and applied

AWS Well-Architected Generative AI Lens (Security)

  • GENSEC03-BP01 Implement control plane and data access monitoring

Learning resources

References

  • AWS SRA for AI — Capability 6, Logging and monitoring