Leo de Moura says AI agents are making formal verification practical
Leo de Moura, creator of the Lean theorem prover and VP at AWS, discusses how AI is transforming formal verification in a recent interview. He notes that while code generation has become easy, verification remains the critical bottleneck. AI models can now perform "autoformalization," translating code from languages like Rust into Lean to enable machine-checked proofs. This capability, which was…
Key points
- AI can now autoformalize code from any language into Lean for machine-checked verification.
- Claude Code used Lean to find and fix real bugs in its own codebase.
However, he emphasizes that AI-generated models of code must be verified for faithfulness to the original source. He points to a recent example where Claude Code found real bugs in its own codebase using Lean, but notes that proving the absence of bugs still requires human oversight of the specifications and models. The conversation also touches on the Navier-Stokes equation proof, where mismatches between informal and formal proofs highlighted the need for better tooling like Verso Blueprint to align natural language and Lean code.
AI, Lean, and the Future of Formal Verification, with Lean Creator Leo de Moura
thedataexchange.media · 29 October 2026
Loading the full article…
This text was published by thedataexchange.media and written by Ben Lorica. It is reproduced here with attribution so you can read it in full; the rights remain with the publisher. Read it at the source ↗
Coverage and discussion
1source- Hacker News discussion · 42 pointsnews.ycombinator.com
The headline, key points and digest above were generated by Digest AI's editorial model from the linked sources. Automated summaries can contain errors: the sources are the record. Spotted a mistake? Tell us. Published by Martin K., who runs Digest AI and handles corrections.
More in Research
All →- Author finds one AI agent follows incorrect reward labels · 1 src
- Flow matching defines a five-stage process for generative AI sampling · 1 src
- Isomorphic Labs aims to design cancer drugs with AI, plans human trials soon · 1 src
- Anthropic's unreleased Claude AI finds critical Linux kernel flaw CVE-2026-98432 · 1 src
- Anthropic opens biology lab with AI agents to find uncharacterized enzyme system · 2 src
Comments
via GitHub Discussions