Mathematicians Validate an AI-Generated Proof of Brennan's Conjecture. Four mathematicians, including 2010 Fields Medalist Stanislav Smirnov, have examined OpenAI's proposed solution to Brennan's conjecture, a longstanding problem in complex analysis dating to 1978. In a manuscript submitted on 16 October 2026 and listed by arXiv on 19 October, Ilia Binder, Oleg Ivrii, István Prause, and Stanislav Smirnov report that the AI-generated argument is correct but unnecessarily lengthy. They identify its essential mathematical ideas, produce a substantially shorter proof, and formulate a stronger result. Brennan's conjecture concerns the integrability of derivatives of conformal maps. For a conformal map g: Ω → 𝔻, where Ω is a simply connected proper plane domain, it predicts ∫_Ω |g′(z)|ᵖ dA(z) < ∞, for 4/3 < p < 4. The researchers also state a stronger bound for the bounded universal integral-means spectrum: B_b(t) < t²/4, for −2 < t < 2, t ≠ 0. The manuscript proves the corresponding non-strict inequality, while the details establishing strictness are deferred to a subsequent paper. The work has not yet undergone peer review. This development illustrates an important relationship between AI-generated mathematics and human mathematical expertise. An AI-produced argument has prompted mathematicians to isolate its central ideas, simplify its reasoning, and investigate a stronger theorem.