Google DeepMind verifies parts of Signal's code with AI proofs
Google DeepMind's AlphaProof Code research produced formal proofs checking selected security properties in Signal's cryptographic software.
By Standard Topic. AI-generated from the reporting listed under Coverage. Published . Updated .
This story is no longer being updated. It reflects reporting up to .
What happened
Google DeepMind researchers used its AlphaProof Code research to generate about 6.5 million lines of machine-checked proofs for Signal's cryptographic code. The proofs verify selected security properties in compiled libsignal code, covering a narrow AArch64 Linux command-line target. A November 11 update expanded machine-code checks to 10 routines and reduced trusted hooks from 37 to 27.
Why it matters
One report says a major flaw stays hidden despite the audit.
Timeline
- Nov 11: A DeepMind update expanded machine-code checks to 10 routines and reduced trusted hooks from 37 to 27.
- Nov 17, 5:25 AM ET: Lean proofs rejected two libsignal builds that had been modified with bugs.
Coverage
- Google DeepMind’s Lean Proofs Reject Two Bug-Modified libsignal Builds (tokenpost.com)
- Signal protocol formal verification, explained (NeoTeo)
- Google AI agents audit Signal’s code and one major flaw stays hidden (Liam 'Akiba' Wright)
More from Technology & AI
- China puts humanoid robots to work serving food and sorting parcels
- Newsom proposes sweeping AI safety rules for Washington
- Google releases Gemini 4 Neon AI model with million-token context
- Microsoft and Intel widen their cloud chip partnership
- Google DeepMind rolls out Genie 4 to select subscribers
- AI helped AMD shorten chip design work, OpenAI says