Policy
See Claude’s rules updated for newer risks
Make it easier to judge what Claude can do in work settings.
Easier access to advanced Claude models for security work
Security professionals can now leverage top Claude models for defensive and authorized offensive work, speeding up vulnerability discovery.
Anthropic expands safer access to advanced cyber features
Qualified defenders can access needed tools more easily.
Strengthen AI safety with embedded external evaluation
Make pre-deployment safety checks more practical.
Anthropic makes it easier to use Claude while keeping company data protected
You can move forward with Claude in enterprise settings while keeping sensitive data better protected.
Safety checks can come earlier for advanced models
Tighter pre-release safety checks for frontier models.