Anthropic releases FlashProve code to verify which model produced a response

Anthropic says it is open-sourcing FlashProve, a cryptographic system that lets developers verify whether a response came from a particular AI model. The company says it is substantially more efficient than prior open-source methods. Potential uses include checking that a model undergoing safety evaluation is the same one served to customers and making chat hijacking harder. Anthropic describes these as applications of the proofs.