Google DeepMind’s AlphaProof Code Verifies Signal’s Cryptographic Core in Four Weeks

What happened

Google DeepMind unveiled AlphaProof Code, an AI system that mathematically proves software meets security specifications, and demonstrated it on Signal’s cryptographic core. Agents built on Gemini Carbon generated more than six million lines of Lean proof code, with roughly one percent of the codebase flagged for human review. The work took four weeks, compared with years typically required, according to the article.

Read next

• More changes in today’s briefingGoogle DeepMind Introduces Gemini 4 Hydrogen for Fast Agentic Workflows

Reported by 5 sources

The story so far

Google DeepMind used AlphaProof Code to formally verify selected cryptographic guarantees in the compiled code underlying the Signal protocol. The four-week proof of concept generated more than six million lines of Lean proofs, which Lean checked automatically; human experts reviewed every specification and examined the roughly one percent of the codebase flagged for a second look. The work found no new vulnerabilities in the portions examined. Tests using two previously fixed bugs produced machine-checked proofs that the older, buggy versions were insecure. DeepMind is releasing its specification and proof to support further formal verification work.