Generative AI driven sensitive data protection
summary
Demonstrates an automated triage that takes Amazon Macie findings, calls Amazon Bedrock (Anthropic Claude 3.7) and SageMaker notebooks, and returns an urgency tag plus recommended Lambda/Step Functions actions; custom identifiers (Medicare, ICD codes) and managed identifiers combine to mark medical PHI as high urgency especially in production accounts. Uses EventBridge to trigger the workflow, queries a CMDB for context, and fetches Macie policy findings; claim is no LLM training required and that the system reduces human escalation by surfacing exact remediation steps.
AI-written summary · about
chapters (10)
- 0:00Introduction to session
- 1:06What is sensitive data
- 3:52Challenges in data protection
- 5:59Healthcare data protection strategy
- 7:01Amazon Macie for data scanning
- 10:33Centralized Macie management
- 13:06Security operations challenges
- 14:51Generative AI in security workflow
- 18:19Demo of AI-driven response
- 24:39Key takeaways and next steps
related talks
19mSecure your healthcare generative AI workloads on Amazon EKS
01h 00mSecurely accelerating generative AI innovation
55mAccelerating auditing and compliance for generative AI on AWS
25mSafeguarding Sensitive Data in the Age of Generative AI
17mSafeguarding sensitive data used in generative AI with RAG
19mKeeping people away from data: A generative AI use case
17mElevating security investigations with generative AI
22mBest practices for data discovery with the new Amazon Macie