FlashProve: a way to check which AI model really answered
Anthropic announced an open-source system that proves cryptographically that an answer came from a given model. It is still research.
A company is working on a way to prove that an answer came from the AI you think it did. It works a bit like signing a message. It is still an experiment and not in use.
Hype check
The noise matches the real change.
Our editorial opinion, based on the sources below. How we score
Your call: hype or real?
Real = it will matter in daily use. Hype = louder than it is. One vote, and it can’t be changed.
Our view and readers’ views are opinions, not statements of fact about a product.
Why it matters
It could stop a cheaper model being swapped in for the one a company names, and let outsiders check.
Watch: like email signing, but for AI models.
The details
- Anthropic announced it in a blog post on November 3, 2026.
- Labs publish a fingerprint for each model; outputs are matched to it and verified cryptographically.
- Fingerprints do not reveal model weights, so closed models can take part.
- The system is light enough to verify on a local device.
- The company published the specification and code under an open license, and has no plans to deploy it yet.
Receipts
One trusted outlet reports it. Not confirmed yet. Open the sources and check us.
Written with the help of AI tools and checked against the sources above. How we work
See a mistake? Tell us and we’ll fix it.
