Summary
Contoso Financial Services started this module with Azure OpenAI models processing sensitive financial data and no threat detection at the AI inference layer. The security team knew they'd enabled Defender for Cloud Apps for Copilot Studio agents, but the Foundry model deployments had no protection—jailbreak attempts, data leakage, and credential theft would go undetected.
You enabled the Defender for AI Services plan on the subscriptions hosting Contoso's AI workloads. Then you configured suspicious prompt evidence so alerts include the context needed for effective triage, and reviewed the plan's optional components for Purview integration and ML model scanning. You then used the Data and AI security dashboard to verify coverage, confirm alerts were flowing, and set up notifications so the SOC team is aware of high-severity AI threats.
In this module, you learned how to:
- Enable the Defender for AI Services plan and configure its components for an Azure subscription
- Investigate AI threat protection alerts in the Defender portal
- Monitor AI security posture using the Data and AI security dashboard in Microsoft Defender for Cloud