EpochLex entry

AI Safety Case

A-I SAYF-tee kays

A structured argument, supported by evidence, that an AI system is acceptably safe within a specified training, deployment, or operating context.