
AI Security2026-11-12
WIRED AI
DeepMind AI tool formally verifies Signal Protocol
Google DeepMind researchers have used an AI tool to formally verify that the Signal encryption protocol works as intended. The result is notable because formal verification is painstaking work that has often been limited to small codebases, while Signal underpins secure messaging for millions of users. DeepMind says the project demonstrates how AI can help prove security properties in real-world software. The team's stated stretch goal is to verify all the software of the world, an ambitious target that highlights both the promise and the limits of automated verification. If such tools can scale, they could change how critical software is built and checked, especially for encryption, authentication, and other security-sensitive systems. Formal verification aims to show that a design satisfies specific mathematical properties, not merely that tests pass. That makes it valuable for protocols where rare flaws can have severe consequences. The Signal work does not mean every implementation is automatically safe, and verification still requires careful definitions and human review. But it offers a credible example of AI assisting with a rigorous form of security assurance. The broader takeaway is that AI is moving into software correctness and security auditing, areas where accuracy and trust are essential.