AI

Anthropic open-sources FlashProve prototype for verifying AI model responses

Anthropic has released FlashProve, an open-source prototype designed to verify which language model generated a particular reply. Developed by Anthropic’s Frontier Red Team with collaborators at EPFL, MIT and StarkWare, it produces a cryptographic proof linking a reply to a specific model. The release includes a prover, two verifiers, formal proofs and a protocol specification, under the Apache 2.0 license. The system addresses the risk that a hosted API router could send a prompt to a different model than the one claimed.

AI-written summary, checked against the cited sources. How summaries work

Image: TechCrunch

What we know

Confirmed by several sources

Reported by one source

  • Anthropic’s Frontier Red Team developed FlashProve with collaborators at EPFL, MIT and StarkWare. — Help Net Security
  • The GitHub release includes a prover, two verifiers, formal proofs and a protocol specification under the Apache 2.0 license. — Help Net Security
  • A hosted API router could route a prompt to a model other than the one represented to the user. — Help Net Security

Coverage 4 publishers

  1. TechmemeFirst report

    Anthropic releases FlashProve, an open-source prototype that cryptographically verifies an AI response came from a given model, not one that a router swapped in

    Roy Rinberg (Anthropic), Alex Allain (Anthropic); Alessandro Chiesa (EPFL, StarkWare), Miranda Christ (MIT, RESI), Rahul Dalal …

  2. Hacker News

    Anthropic releases FlashProve, an open-source prototype that cryptographically verifies an AI response came from a given model, not one that a router swapped in

    Roy Rinberg (Anthropic), Alex Allain (Anthropic); Alessandro Chiesa (EPFL, StarkWare), Miranda Christ (MIT, RESI), Rahul Dalal …

  3. TechCrunch

    Anthropic is working on a system to prove AI models’ identity

    Called FlashProve, the system would let anyone check that an AI response really came from the model it claims to be from.

  4. Help Net Security

    Anthropic open-sources FlashProve, its provable inference prototype for MoE models

    Anthropic’s Frontier Red Team, with collaborators at EPFL, MIT and StarkWare, built FlashProve, a prototype that produces a cryptographic proof that one specific language model generated one specific reply. The prover, two verifiers, the…

Earliest report first. Articles stay on their publishers’ sites; each link opens the original.