AlphaProof Code
Points and comments are a snapshot, not live.
DeepMind's AlphaProof Code formally verified Signal's cryptographic protocol for the first time.
Google DeepMind introduced AlphaProof Code, an agentic framework that formally verified key security properties of the Signal protocol's code. Human experts and AI agents translated security goals into mathematical specifications, and agents powered by Gemini Carbon generated over six million lines of Lean proofs. Only one percent of the codebase required human review. This marks the first AI-driven formal verification of a widely deployed cryptographic protocol at this depth.
The work proves the protocol's implementation is mathematically correct, a standard far beyond typical security testing.
What commenters are saying
Commenters praised the achievement, noting Signal's code is now formally verified for cryptographic security, a rare bar. Several noted the verification covers only the encryption protocol, not the full app or OS, so other attack surfaces remain. A debate emerged about US government ties to Signal and the EFF, with one commenter alleging they are military programs larping as civilian institutions. Others discussed the halting problem's theoretical limits, explaining that Lean's type system bypasses it in practice.