Security, compliance and governance for AI cheat sheet
All 6 AIF-C01 flashcards for Security, compliance and governance for AI on one page, grouped by exam objective. Print it or save it as a PDF for a last-minute read.
Explain methods to secure AI systems
- Shared responsibility on Amazon Bedrock: who secures what?
- AWS secures the cloud itself: the infrastructure Bedrock runs on. You secure what you do in it: your data, who has access and how it's configured.
- Why connect to Amazon Bedrock through AWS PrivateLink?
- Your VPC reaches Bedrock over a private connection. No internet gateway or public IP addresses are needed.
- How does a contextual grounding check reduce hallucinations?
- It checks each response against the source documents and the question. Answers the source doesn't support, or that miss the question, are blocked.
- What does AgentCore Identity do?
- It gives AI agents their own identity and manages their credentials. Agents can then access AWS and third-party services on a user's behalf, with an audit trail.
Recognize governance and compliance regulations for AI systems
- Can model providers see your prompts on Amazon Bedrock?
- No. Bedrock runs each provider's model in AWS-owned accounts the provider can't access, so they never see your prompts, responses or logs.
- What is the Generative AI Security Scoping Matrix for?
- It sorts a GenAI workload into one of five scopes, from using a public app to training your own model. The scope shows which security controls to focus on.