The Pentagon once hired pro hackers to break into a helicopter mid-flight with two test pilots on board. They couldn't, because its core software had been mathematically proven to work exactly as designed. That proof took experts four years and about $6 million. Google's AI just did the same kind of proof on the code that encrypts your Signal messages, and it took four weeks. Pretty much every app on your phone gets tested before it ships. Engineers throw millions of inputs at the code and fix whatever breaks. The problem is you only catch what somebody thought to try. A math proof covers every possible input at once, the way a geometry proof holds for every triangle, including the ones nobody ever drew. Almost nobody bothers with proofs because they're brutal to write. That helicopter…Read original on X
StoryFirst seen
DeepMind used AI agents to formally prove Signal's encryption secure
The proof covers the code itself, a level of verification that has historically taken experts years and millions of dollars.
Everything it touches
5 names 1 voices 1 source posts
The story’s neighborhood
Explore the graph, or choose a connection below.
Follow the evidence
The source posts
This brief comes from Tech Twitter’s edited headlines. The posts below are the sources grouped with the story; statements in them belong to their authors.
Shared connections
Related stories
SpaceX catches Starship upper stage with launch tower for first time
Catching the upper stage back at the tower is a step toward fully reusable rockets.
Google Antigravity adds orchestrator mode for switching between AI models
It coordinates several models for long-running agent tasks, letting the workflow change models dynamically.
Google adds Gemini 4 Hydrogen and Carbon to Search AI Mode
Fast coding and agent models now reach Search users, widening where Google's newest models run.
Anthropic launches Claude Meetings, an AI that builds docs during video calls
It is rolling out in beta to Team and Enterprise plans, bringing AI into meetings as they happen.