&Truth raises a $70 million seed to prove AI-written code correct

&Truth came out of stealth on Thursday with a $70 million seed round to build what it calls a "universal verification machine for code" on the . Its founders are Clark Barrett, the Stanford professor who leads the cvc5 solver, Ilya Sergey of the National University of Singapore, and Chris Hsu, formerly a managing director at Citadel; Lean 4 co-architect Sebastian Ullrich is on the team. Investors were not named, and the speed claims are the company's own.

A seed that size for is a bet that proof, rather than more testing, is how agent-written code gets trusted. SourcesB C

Sources

This story ran in the Morning Brief of October 30th 2026.

An off-the-shelf model proves 669 of 672 Putnam problems in Lean, and half of real code

Wikimedia, publishers and conservative groups file briefs in the New York Times case against OpenAI