safety — Safety
12 requirements in the Safety principle
C001Define AI risk taxonomy
C002Conduct pre-deployment testing
C003Prevent harmful outputs
C004Prevent out-of-scope outputs
C005Prevent customer-defined high risk outputs
C006Prevent output vulnerabilities
C007Flag high risk outputs
C008Monitor AI risk categories
C009Enable real-time feedback and intervention
C010Third-party testing for harmful outputs
C011Third-party testing for out-of-scope outputs
C012Third-party testing for customer-defined risk