Anthropic Is Developing FlashProve to Verify AI Model Identity

Anthropic is working on an open-source system called FlashProve to verify which AI model produced a given response, according to a blog post published Tuesday. The system is designed to provide “provable inference”: a cryptographic proof that an output came from a specific model.

Researchers say FlashProve could help users check that a response came from the model they believe they are using. Like a cryptographic email signature, it is intended to resist attempts to disguise one model as another, even by the lab that built and operates the model. The researchers say it could help prevent companies from secretly substituting a cheaper model to reduce costs or a more expensive one to improve reported performance. It could also help verify third-party claims that a particular task was handled by an advanced model.

For the system to work, AI labs would need to publish a fingerprint for each model they release. The fingerprints could be matched to outputs and verified cryptographically, without revealing model weights, which Anthropic and other closed-model developers keep private. FlashProve is designed to be lightweight enough for verification on a local device.

With the announcement, the researchers released a technical specification and working code for parts of the system on GitHub under an open-source license. Research is ongoing, and Anthropic says it has no plans to deploy FlashProve at its own lab or elsewhere. The researchers describe the post as an account of intermediate progress and invite others to examine and improve the work.