Google DeepMind researchers used AI agents to formally verify key security properties of the actual code behind the Signal protocol, the technology that protects billions of private conversations in apps like Signal and Google Messages, the company said.[1][2]
This is the first time any method, with or without AI, has formally verified the implementation of a widely deployed cryptographic protocol, the researchers wrote.[1][2] According to Google DeepMind, the work was done using AlphaProof Code, a novel agentic framework that combines human expertise with AI agents to prove key security properties of the actual code underlying the Signal protocol.[1]
The agents, powered by Gemini Carbon, generated over six million lines of Lean, a formal proof language, to verify the code, a task that would typically take expert teams years but was completed in four weeks, according to Google DeepMind.[1][3]
To test the system, researchers reintroduced two previously fixed bugs and confirmed AlphaProof Code correctly proved the resulting code insecure.[1][2]