Anthropic is developing an open-source system called FlashProve to verify which AI model produced a given output. The system delivers what researchers call provable inference, a cryptographic proof that provides a mathematical guarantee the response came from a specific model. It works like cryptographic email signatures and is designed to resist impostors, including cases where a lab secretly substitutes a cheaper or more capable model. AI labs would publish a fingerprint of each model that can be matched to outputs and verified without revealing the model’s weights. Verification is intended to run on a local device. Anthropic has released a technical specification and working code on GitHub under an open-source license. The company describes the work as intermediate research progress with no current plans for deployment at Anthropic or other labs, inviting others to build on the approach.
Anthropic is developing an open-source system called FlashProve to verify which AI model produced a given output. The system delivers what researchers call provable inference, a cryptographic proof that provides a mathematical guarantee the response came from a specific model. It works like cryptographic email signatures and is designed to resist impostors, including cases where a lab secretly substitutes a cheaper or more capable model. AI labs would publish a fingerprint of each model that can be matched to outputs and verified without revealing the model’s weights. Verification is intended to run on a local device. Anthropic has released a technical specification and working code on GitHub under an open-source license. The company describes the work as intermediate research progress with no current plans for deployment at Anthropic or other labs, inviting others to build on the approach.