Cursor Cyber Safeguards & Anthropic CVP for Security Teams

Blocks on sensitive but legitimate security tasks (pentest planning, malware analysis write-ups, defensive tooling) come from Anthropic's API safeguards, not Cursor policy. Cursor surfaces CVP application and model toggles in the dashboard, but Anthropic sets program terms, approval and data handling for cyber-verified models.

Cyber Safeguards apply at Organization Group level (Organization → Groups), not billing groups or SCIMSystem for Cross-domain Identity Management. A standard for automatically creating and removing user accounts when people join or leave. directory groups. Only members of the approved group get cyber-verified model access; the rest of the org keeps normal Privacy ModeCursor's setting that routes requests under zero-data-retention terms so providers don't store or train on your code. behavior.

1. 1Open Organization → Groups and create or select a dedicated group (for example Security or CVP).
2. 2Add only the members who need cyber-verified model access.
3. 3In Cyber Safeguards Models Settings, click Apply and accept Anthropic's terms as an authorized representative.
4. 4Wait for Anthropic approval; Cursor rechecks status about every two hours and shows Approved when cleared.
5. 5Enable the Cyber toggle for Opus 4.7 or 4.8 on that group.

Cursor documents Cyber mode for Opus 4.7 and Opus 4.8 only. Pricing matches the base model rates in the models and pricing doc. If approval is pending, check back after Anthropic completes review; instant approval is not guaranteed.