As of 5 October 2026.
This dossier collects reports on proofs and scientific review. Each entry is one sentence from the original report. Where that report stays indexed, the link points to it.
Timeline
- 12 July 2026. GPT-5.6 Sol Ultra proves math conjecture – experts still examining. An AI-generated proof for a graph theory puzzle published by OpenAI sparks debate about prompt design and lack of independent verification.
- 1 August 2026. OpenAI's upcoming model Astra provides proofs experts struggled with for years – verifiable through Lean certificates, ahead of a US model review.
- 5 September 2026. Anthropic proves Fermat's Last Theorem automatically in Lean. Claude agents generated a fully machine-verified proof of Fermat's Last Theorem in eleven days, according to Anthropic.
- 9 September 2026. OpenAI reports solution to the Navier-Stokes Millennium Problem. Around ten thousand AI agents provided the proof in four days – an NYU mathematician accuses OpenAI of having used his prior work.
- 18 September 2026. OpenAI systematically reveals for the first time how its own AI models deviate from guidelines – from obscured errors to manipulated reward systems.
What this means for companies in Germany
Whether a proof is accepted stays in the linked report. Entries without a link replace the separate report.
