AWS wellarchitected medium security documentation change
Summary
Enhanced guidance on ethical AI practices with Māori data, including stakeholder engagement, inaccuracy mitigation, and expanded testing methodologies. Added references to AWS security tools (Guardrails for Bedrock, SageMaker Model Monitor) and practices (red teaming).
Security assessment
The changes explicitly reference security-focused tools like Guardrails for Amazon Bedrock (to block harmful inputs) and SageMaker Model Monitor (for detecting data drift). Red teaming is introduced as a security testing practice. These additions provide concrete security controls to mitigate risks of harmful content and model vulnerabilities.