Google DeepMind verifies Signal's encryption with AI

DeepMind proves Signal's encryption works

Google DeepMind researchers used artificial intelligence to formally verify key cryptographic guarantees in the Signal protocol, mathematically proving it functions as intended.

The work protects billions of messages

The verification validates the cryptography protecting massive quantities of communications on Signal app, Google Messages and WhatsApp, serving as independent validation.

Four weeks to prove the protocol

Conducted as a proof-of-concept project in just four weeks, the work demonstrates Google DeepMind's formal verification capabilities using a new AI framework called AlphaProof Code.

What formal verification actually does

Formal verification uses mathematics to check and validate that software functions as intended, meaning it does not have critical vulnerabilities. Before AI, this was incredibly time-consuming.

No new vulnerabilities found in Signal

The DeepMind analysis did not find any new vulnerabilities in the portions of Signal protocol it examined. When tested on older versions with known flaws, it correctly identified them.

The scaling question remains open

The open question across tech is whether AI-automated formal verification can truly scale and do so at workable cost. DeepMind conducted the Signal analysis free as a research project.

“If you can now verify the correctness of these implementations, then you can rest easy.”

Human-AI collaboration is the middle ground

DeepMind emphasizes a human-AI collaboration process where agents contribute and self-correct at almost every phase while still seeking human input and review at key junctures.

“We don't want this sort of vibe world where AI is generating the goals and AI is generating the proofs and basically taking tests that are self-written.”