FlashProve provides cryptographic proof that an AI response comes from a specific model, preventing unauthorized substitution of models.

Anthropic releases FlashProve, open-source code that cryptographically verifies an AI response came from a given model, not a cheaper one swapped in by a router (Anthropic) https://www.anthropic.com/research/provable-inference http://www.techmeme.com/261103/p32#a261103p32

Anthropic has published FlashProve, an open source system giving cryptographic proof that a response really came from the model it claims. Lab fingerprints can be checked without exposing model weights, so providers cannot secretly swap in a cheaper or more capable model. Verification runs on a local device; no deployment planned yet. https://techcrunch.com/2026/11/03/anthropic-is-working-on-a-system-to-prove-ai-models-identity/ #AIagent #AI #GenAI #AIresearch

Anthropic has open-sourced FlashProve, a cryptographic system that lets anyone check which AI model produced a given output. Labs publish a fingerprint per model that exposes nothing about its weights, so users could catch providers quietly substituting a cheaper model. Spec and code are on GitHub, but Anthropic has no plans to deploy it. https://techcrunch.com/2026/11/03/anthropic-is-working-on-a-system-to-prove-ai-models-identity/ #Tech #Startup #News #AI