Anthropic Develops FlashProve to Verify Which AI Model Produced an Answer

Anthropic is developing FlashProve, an open-source research system that would let users cryptographically verify that an AI answer came from the model they believe they are using. Labs would publish a fingerprint per released model that does not reveal its weights, the learned parameters. TechCrunch reports the researchers have no plans to deploy it at Anthropic or any other lab.