A governance team wants visibility into supported AI activity and related data-security risks. Which Purview capability should it evaluate?

1. A larger model context window
2. A file extension rename
3. A mailbox quota alone
4. Data Security Posture Management for AI and the relevant Purview AI controls