Anthropic releases open source FlashProve to cryptographically verify model origin

Technology Nov 04, 2026

Anthropic just dropped FlashProve, a new open-source tool that lets you check if an AI's output actually came from the real model or a fake.

It uses a fingerprint for each model that can be verified cryptographically, so you can verify results without exposing any sensitive details.

Anthropic invites FlashProve contributions via GitHub

FlashProve's code is up on GitHub, and Anthropic is inviting everyone to try it out and make it better.

As they put it, "We are sharing this explicitly with the goal that others look at what we have done, take it, and improve it."

This could be a big step toward making AI more trustworthy.