SafetyPage

Lab research on the web · Blog post · Anthropic

LLMs are making formal software verification feasible at a much larger scale

Read at anthropic.com ↗Read at sourceAll rights reserved

Outline · 57 paragraphs

  1. Introduction¶3
  2. Turning tests into proofs¶12
  3. Finding and fixing bugs¶25
  4. Proving more than memory safety¶30
  5. Other formal verification case studies¶33
  6. This is just the beginning¶39
  7. Newer models need fewer tokens¶41
  8. Proofs can be reused¶43
  9. Limitations¶49
  10. Conclusion¶55