# Anthropic releases open source FlashProve system to verify which AI model produced an output

published: 2026-11-04T02:05:25.574Z
topic: infrastructure
source: https://techcrunch.com/2026/11/03/anthropic-is-working-on-a-system-to-prove-ai-models-identity/
url: https://aiagentnews.io/stories/anthropic-releases-open-source-flashprove-system-to-verify-which-ai-model

Anthropic is working on an open source system called FlashProve that uses cryptographic proofs to show a given output came from a specific AI model. The researchers have published a technical specification and working code on GitHub, though they say the research is ongoing and there are no plans to deploy it at Anthropic or other labs.

The system is meant to provide what researchers call provable inference, a cryptographic proof that an output came from a particular model. It would let users confirm that an answer really came from the model they believe they are using, and it would guard against an impostor disguising one model as another, even when the impostor is the lab that runs the model.

To work, labs would publish a fingerprint of each model they release, which can later be matched to specific outputs. The fingerprints do not reveal model weights, so closed-model makers could participate without exposing their most valuable assets. The system is designed to be lightweight enough for verification on a local device.

The researchers say FlashProve could prevent a provider from secretly swapping in a cheaper or more expensive model, and could let third parties verify that a task was done by an advanced model. The researchers describe the work as intermediate progress and say they hope others will take it and improve it.

## Why it matters if you build agents

Agent developers who depend on a specific hosted model may eventually be able to check that outputs came from the model they paid for. The system is still research, though, and no deployment is planned at Anthropic or other labs, so it should not be relied on in production yet.