Google DeepMind researchers used AlphaProof Code to formally verify parts of the Signal protocol in four weeks.
The AI agents wrote more than six million lines of proof code in the Lean language and flagged only about one percent of the code for human review.
The system correctly identified two deliberately planted bugs in older Signal versions and the researchers released the specification and proof to aid future verification work. DeepMind plans to extend the method to critical infrastructure like power grids and medical devices. Signal CTO Ehren Kret stated the verification strengthens trust as AI threats grow.
