DeepMind formally verifies parts of the Signal protocol with AlphaProof Code
Original title: Google DeepMind just mathematically verified key parts of Signal
A four-week proof-of-concept found no new flaws in the portions checked and correctly flagged known bugs in older Signal versions.
Google DeepMind researchers used an agentic framework called AlphaProof Code to mathematically prove several cryptographic guarantees of the Signal protocol. The work, done as a proof of concept in four weeks, did not cover the entire Signal mobile app and found no new vulnerabilities in the portions examined. When run against older Signal versions with known, long-fixed flaws, the system identified them and showed the protocol could not be formally verified. Researchers are releasing the specification and proof.
Unconfirmed: Researchers say it is unclear how practical AI-aided formal verification will be at scale.
- The analysis was completed in four weeks as a proof of concept.
- It did not formally verify the entire Signal mobile app.
- AlphaProof Code flagged known flaws in older Signal versions.
- The specification and proof are being released for other verification work.
- Signal, Google Messages, and WhatsApp rely on the protocol.
- Vendors
- Products
- SignalGoogle MessagesWhatsApp
- Organizations
- Google DeepMind
- AI models
- AlphaProof Code
From the original article
2 min read at arstechnica.com
The four-week project hints that AI could make formal verification far more practical.
By Lily Hay Newman, wired.com · arstechnica.com
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.