🔐DeepMind Uses AI Agents to Formally Verify Parts of Signal's Code in One Month
Your WhatsApp messages just got a little more provable
TL;DR
Google DeepMind used AI agents to mathematically prove parts of Signal's encryption code do what they claim, in about a month. WhatsApp and Google Messages share the same protocol, so the proofs matter beyond Signal. Open-source code is coming for anyone doing formal verification.
Google DeepMind just used AI to mathematically prove that parts of Signal's encryption code do what they say, with no errors. The whole job took about a month. Expert humans usually need far longer to build a proof like this by hand. If you ship security-sensitive code, this is the part to care about. The workflow was split: humans worked with agents to decide what needed proving, then multiple agents executed the proofs. The agents also audited the proof and flagged spots that needed a second human look. That's a template you can actually copy. The catch? DeepMind did not verify all of Signal's code, only portions. There are no plans to turn this into a consumer product. The researchers are open-sourcing their code, and WhatsApp and Google Messages use the same encryption as Signal, so the groundwork reaches far beyond one app.

Key Points
DeepMind's verification took roughly one month, far faster than the normal timeline for experts to build equivalent manual proofs of encryption code.
Keep reading — it's free
Enter your email to keep reading — plus the best of AI & tech, daily. Free, forever.
Already a member? Sign in
Comments
Be the first to comment