Anthropic Developing FlashProve System to Verify AI Model Identity

As reported by TechFlow on November 4, citing TechCrunch, Anthropic is developing the open-source system FlashProve, designed to verify whether an AI output comes from a specified model through cryptographic proofs.

The system requires model providers to publish model “fingerprints” for users to match and verify against outputs, while the fingerprints do not leak model weights. Researchers said this could help prevent service providers from secretly swapping models and verify third-party claims about using specific models.

Researchers have published the technical specifications and some system code, but the research is ongoing, and there are currently no plans to deploy the system at Anthropic or other labs.

Source: TechFlow