Source: Wired
Formal verification—mathematically proving code correctness rather than testing it—has been theoretically sound but economically prohibitive for decades, confined to safety-critical systems like avionics. The emergence of AI-assisted verification startups like &Truth, backed by $70M seed rounds, reflects a structural bet that LLMs and automated reasoning can reduce the engineering overhead enough to move verification into mainstream software development, where unverified code currently dominates because speed-to-market beats certainty. Success would lower the cost of verification enough to justify it in ordinary software; failure would expose how AI-generated code makes verification more necessary, not less.