Question

Contoso has deployed Microsoft 365 Copilot and enabled DSPM for AI. The security team reports that employees are pasting content labeled 'Highly Confidential' into third-party generative AI sites such as public ChatGPT, and they also want to reduce the risk of users prompting Copilot with sensitive data. Which combination of Purview DSPM for AI capabilities should you configure to address BOTH concerns with the least administrative effort?

Accepted Answer

Use the DSPM for AI recommendation to create a DLP policy that restricts sensitive content in AI prompts, and apply the 'Fortify your data security for AI' policy that detects sensitive interactions in Copilot.

Answer

Deploy an Endpoint DLP policy alone to block all AI websites for every user, and disable Copilot for the tenant.

Answer

Configure only an Insider Risk Management policy using the 'Data leaks' template to capture browsing to AI sites.

Answer

Enable Audit (Premium) logging for Copilot interactions and rely on Activity explorer to manually review and remediate risky prompts each week.