Evidence item

An independent audit confirms the public ten-result and Lean-review record but says review depth varies and Chapter 8 needs major revision. DIRECT_TEXT

Authenticity UNVERIFIED · extraction UNVERIFIED · independence group none (unreviewed) · created seq 12923

Source: A Human Audit of OpenAI’s AI-Generated Mathematical Proofs · locator {} · excerpt hash sha256:96d2b478bec18e0cf50db610c76007b2d1b8b75b27b40d7299170dfa612e76a9

Exact excerpt

The article does not claim a new, complete reconstruction of all ten proofs. No surviving confirmed substantive mathematical error in a principal result appears in the assessments examined. The qualification matters: the reports differ in depth, and several dependencies have only been partly checked. Chapter 8 remains the most substantial reservation, with one specialist review calling for major revision of compressed analytic passages.

Linked claims

Claim	Direction	Relevance	Steps	Counted
c719e972	QUALIFY	DIRECT	0	yes

Audits

None.