Urgent.News

What's breaking now, across thousands of outlets.

AI

Anthropic AI ‘formalizes’ proof of Fermat’s last theorem in just 11 days

Nature, Published online: 07 September 2026; doi:10.1038/d41586-026-02822-9 Claude produced a 13-million-line, computer-checked proof of the famed conjecture — a major milestone in mathematics.

Anthropic AI has formally verified a proof of Fermat's Last Theorem in just 11 days, using an advanced prototype of its artificial-intelligence chatbot Claude. The AI's achievement has astonished researchers, who are amazed at the rapid progress of AI's mathematical capabilities. Claude's feat is particularly impressive, as it completed a project that would have taken humans around 10 years to finish.

This breakthrough suggests that AI could play a significant role in checking the work of mathematicians and even in producing new mathematical reasoning. As AI's mathematical skills continue to improve, it may soon be able to scrutinize the entire library of mathematical knowledge, potentially finding errors in well-known results.

Kevin Buzzard, a mathematician at Imperial College London, describes the AI's formalization of proofs as a whole new level of complexity compared to previous milestones, such as its certification of Maryna Viazovska's work on sphere packing.

Written by urgent.news from Nature's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.

Read the original at nature.com →

More in AI

19 Audit Nags in One Night: Making a Claude Code Stop Hook Detect Unattended Sessions

My autonomous setup earns its keep precisely because Claude Code starts on its own and finishes on its own. The thing that nearly broke that setup was Claude Code itself.

  • Claude Code's autonomous notifications persist without human presence.
  • Self-audit script fails in automated pipeline without human user.
  • Audit mechanism applied only to human-initiated sessions to prevent interference.

More from Monday 7 September →