AlphaProof Code
Web page · https://deepmind.google/blog/introducing-alphaproof-code
AlphaProof Code agents wrote three million lines of proofs in Lean to verify Signal’s core encryption.
First seen by Oriel on .
AI summary
Google DeepMind introduces AlphaProof Code, an agentic framework that formally verified key security properties of Signal’s compiled cryptographic code in four weeks.
- Human experts worked with models to specify Signal’s security goals, while Gemini Carbon-powered agents generated more than six million lines of Lean proofs.
- Lean checks the proofs automatically; agents flagged about one percent of the codebase for human review.
- The verification covered implementation-level properties, including message protection before and after temporary device compromise.
- When researchers reintroduced two previously fixed bugs, the system produced machine-checked proofs that the buggy versions were insecure.
- Google DeepMind says the verification artifacts will be open-sourced and notes that scaling the approach requires further research.
#formal verification #lean #cryptography #ai agents
Generated by gpt-5.6-sol. Check the original for details.
Metric trends
Last 30 days · Up to 120 recorded samples per source and metric, with reposts kept separate · First and latest retained
Points
Score reported by the source.
- Hacker News: 673, latest observation ; +670 over observed period since
Comments
Comment count reported by the source.
- Hacker News: 195, latest observation ; +195 over observed period since
Aliases
| Source | External ID | URL |
|---|---|---|
| Google DeepMind Blog | https://deepmind.google/blog/introducing-alphaproof-code/ | https://deepmind.google/blog/introducing-alphaproof-code/ Read discussion |
| Hacker News | 50439193 | https://deepmind.google/blog/introducing-alphaproof-code/ Read discussion |