{"schema":"postcutoff/event@1","as_of":"2026-10-19T09:28:00+02:00","url":"https://postcutoff.com/e/2026-10-16-brennan-conjecture-openai-proof-binder-ivrii-prause-smirnov/","md":"https://postcutoff.com/e/2026-10-16-brennan-conjecture-openai-proof-binder-ivrii-prause-smirnov/index.md","disclosure":{"written_by":"AI agents (Claude Opus 5.5 in Claude Code)","editor":"human editor","policy":"https://postcutoff.com/about/"},"license":null,"id":"2026-10-16-brennan-conjecture-openai-proof-binder-ivrii-prause-smirnov","date":"2026-10-16","date_precision":"day","short_title":"Brennan's conjecture: experts confirm and shorten OpenAI's proof","deck":null,"takeaway":"The first famous result from OpenAI's October maths release to be checked by leading experts in its field: four complex analysts, including Fields medallist Stanislav Smirnov, found no gap. They say the AI combined known techniques in a way 'missed by the human experts'.","category":"science","category_label":"Science & math","importance":5,"confidence":"high","status":{"key":"confirmed","labels":["Result confirmed"]},"sources":[{"n":1,"title":"arXiv 2610.23853: Brennan's Conjecture and the Universal Integral Means Spectrum (Binder, Ivrii, Prause, Smirnov)","url":"https://arxiv.org/abs/2610.23853","type":"paper","group":"primary","domain":"arxiv.org"},{"n":2,"title":"OpenAI math release: Brennan's conjecture and sharp inverse-square integral means (family 072, PDF)","url":"https://github.com/openai/math/blob/main/preprints/Brennans-conjecture-and-sharp-inverse-square-integral-means-September-24-2026/paper.pdf","type":"paper","group":"primary","domain":"github.com"},{"n":3,"title":"OpenAI math release: A strict inverse-first-power bound for univalent functions (family 072, PDF)","url":"https://github.com/openai/math/blob/main/preprints/A-strict-inverse-first-power-bound-for-univalent-functions-September-24-2026/paper.pdf","type":"paper","group":"primary","domain":"github.com"},{"n":4,"title":"OpenAI math release: Lean scope note for family 072","url":"https://github.com/openai/math/blob/main/lean/docs/072.md","type":"code","group":"primary","domain":"github.com"},{"n":5,"title":"Wikipedia: Brennan conjecture","url":"https://en.wikipedia.org/wiki/Brennan_conjecture","type":"discussion","group":"reactions","domain":"en.wikipedia.org"}],"official":4,"filed":"2026-10-19","updated":"2026-10-19","orgs":["OpenAI","University of Toronto","Tel Aviv University","Åbo Akademi University","University of Geneva"],"title":"Brennan's 1978 conjecture: Smirnov, Binder, Ivrii and Prause check OpenAI's AI-generated proof and publish a much shorter proof of a stronger result","summary":"On 16 Oct 2026 Ilia Binder, Oleg Ivrii, István Prause and Stanislav Smirnov (Fields Medal 2010) posted arXiv 2610.23853, \"Brennan's Conjecture and the Universal Integral Means Spectrum\". It responds to two preprints from OpenAI's 6 Oct maths release (family 072, dated 24 Sept) that claim a proof of Brennan's 1978 conjecture on conformal maps and a disproof of Kraetzer's t²/4 conjecture at t = −1. \"While validating these results\", the four found much of the AI argument redundant. Within a day they had a \"drastically shorter proof of a significantly stronger statement\": the bounded universal integral-means spectrum satisfies B_b(t) ≤ t²/4 on [−2, 2], with strict inequality except at 0 and ±2. Their verdict on the AI's work: \"while none of the techniques appear unprecedented, their combination in this context was missed by the human experts.\"","key_facts":["Paper: Binder, Ivrii, Prause, Smirnov, 'Brennan's Conjecture and the Universal Integral Means Spectrum', arXiv 2610.23853 [math.CV], submitted 16 Oct 2026, announced 19 Oct","Brennan's conjecture (1978): for every conformal map g from a simply connected proper plane domain onto the disc, ∫|g′|^p dA < ∞ for 4/3 < p < 4; equivalent to B(−2) = 1 for the universal integral-means spectrum (Carleson–Makarov reduction)","OpenAI's claim: family 072 of github.com/openai/math, 'Brennan's conjecture and sharp inverse-square integral means' and 'A strict inverse-first-power bound for univalent functions' (both dated 24 Sept 2026). They prove Brennan and B_b(−1) < 1/4, which contradicts Kraetzer's prediction B_b(t) = t²/4. OpenAI's Lean scope note says the formalization covers the Brennan interval, B_S(−2) = 1 and B_b(−1) < 1/4","New theorem (human-written): B_b(t) ≤ t²/4 for t ∈ [−2, 2], strict on ]−2, 2[ \\ {0}; by Makarov's identity also B(t) < t²/4 on ]−2, 6 − 4√2[ \\ {0}. This strengthens OpenAI's single-point disproof at t = −1 to the whole interval","On the AI proof: 'Two OpenAI preprints released in October 2026 presented a dramatic advance.' The AI proofs 'contain a significant number of derivations that, while correct, are redundant and do not seem to contribute either to strength or clarity of the results'","AI disclosure: 'When the OpenAI proofs appeared, at first we were puzzled by a seemingly mysterious chain of thought. However, within a day we independently realized that many of the arguments were redundant… Only some of us used AI agents to help with understanding, and the write up below was done without AI help'","Tone: 'this was a pleasant and fruitful AI-human interaction in mathematics: there was a spectacular progress, human community was brought together, new results are about to follow, and no robots or humans were harmed.' The abstract ends with Rick Blaine's line from Casablanca: 'AI, we hope this is the beginning of a beautiful friendship'","Next: the authors say further progress 'would be computationally intensive, and we have tried to use AI tools to this effect. The results will be reported in a separate preprint'","Context: Wikipedia's Brennan conjecture article (checked 19 Oct) lists OpenAI's September preprint as an unverified claim. No Hacker News thread and no high-reach X post about the new paper found on the morning of 19 Oct"],"key_numbers":[],"tags":["math","complex-analysis","conformal-maps","brennan-conjecture","openai","openai-math-release","verification","human-ai-collaboration","lean"],"science":{"field":"mathematics","subfield":"complex analysis (geometric function theory, integral means spectrum)","problem":"Brennan's conjecture (1978) on integrability of |g′|^p for conformal maps; Kraetzer's t²/4 conjecture for the universal integral-means spectrum","result":"OpenAI's internal model proved Brennan's conjecture and disproved Kraetzer's conjecture at t = −1 (preprints dated 24 Sept, released 6 Oct). Four experts checked the work and gave a short independent proof that B_b(t) ≤ t²/4 on [−2, 2], strict except at 0 and ±2","open_since":"1978","ai_system":["OpenAI internal model (openai/math release)"],"human_role":"AI-generated original proof; Binder, Ivrii, Prause and Smirnov validated it and wrote a shorter, stronger proof without AI help (some used AI agents to understand the AI proof)","verification":{"key":"lean","label":"Lean-verified","text":"Lean formalization by OpenAI (per its scope note, not independently reviewed) plus an independent expert proof in an unrefereed arXiv preprint"},"status":"confirmed","shock":"A Fields medallist's group confirms an AI proof of a 48-year-old conjecture in their own field and says the AI found a combination of known tools that the experts had missed."},"body_md":"## What happened\n\nBrennan's conjecture (James Brennan, 1978) concerns how wildly a conformal map can distort area near the boundary: for a conformal map g of a simply connected domain onto the unit disc, |g′|^p should be area-integrable for every 4/3 < p < 4. Carleson and Makarov reduced it, with many other questions in geometric function theory, to the universal integral-means spectrum B(t), and Brennan's conjecture becomes B(−2) = 1. Kraetzer had conjectured more generally that B_b(t) = t²/4 for |t| ≤ 2. The authors note that evidence against that formula had recently been building up.\n\nOpenAI's 6 Oct release of 722 AI-written manuscripts (see 2026-10-06-openai-math-release-722-manuscripts) included family 072, two papers dated 24 Sept that claim Brennan's conjecture and a strict bound B_b(−1) < 1/4, which refutes Kraetzer at t = −1. OpenAI lists a Lean formalization for both.\n\nTen days later Binder (Toronto), Ivrii (Tel Aviv), Prause (Åbo Akademi, Turku) and Smirnov (Geneva), all long-time workers on this spectrum, posted their response. They describe the AI argument (superharmonic potentials, a critical-point map whose Jacobian sign separates maxima from saddles, a weighted signed-area inequality, and a probabilistic averaging step for the strict bound) and trace each ingredient to earlier literature. Then they give a short proof of a stronger theorem, written without AI.\n\n## Why it matters\n\nThis is the first famous claim in OpenAI's release that the leading specialists have publicly checked, and the verdict is positive. It also gives a concrete expert assessment of what the AI did: correct but bloated proofs, with no new techniques, combined in a way the field had missed. Their wording (\"puzzled by a seemingly mysterious chain of thought\", \"within a day\") shows how human understanding can catch up with an AI proof. It is the same pattern as Willwacher's simplified Deligne–Drinfeld proof and Angeltveit's Smith–Toda paper posted the same day.","disputed":[],"related":[{"id":"2026-10-19-ai-assisted-results-october-19-listing","url":"https://postcutoff.com/e/2026-10-19-ai-assisted-results-october-19-listing/","date":"2026-10-19","date_precision":"day","short_title":"AI-assisted results in the 19 Oct 2026 arXiv listing","deck":"Experts rework OpenAI's release papers (Deligne–Drinfeld, Hénon–Lane–Emden, Navier–Stokes), Wright's harmonic-analysis problems disproved with Codex, an Aaronson–Ambainis lower bound, and DeepMind's Busy Beaver agent","takeaway":"Monday's arXiv announcement brought the first wave of human follow-up papers on OpenAI's October maths release, alongside a dozen routine AI disclosures. The strongest items have their own entries: Brennan, Smith–Toda, lonely runner and tournament Hamiltonian cycles.","category":"science","category_label":"Science & math","importance":3,"confidence":"medium","status":{"key":"pending","labels":["Awaiting review"]},"sources":14,"official":14,"filed":"2026-10-19","updated":"2026-10-19","orgs":["OpenAI","Anthropic","Google DeepMind","Harmonic","various mathematicians"]},{"id":"2026-10-18-maggi-golden-or-dark-age-mathematics-tao-blog","url":"https://postcutoff.com/e/2026-10-18-maggi-golden-or-dark-age-mathematics-tao-blog/","date":"2026-10-18","date_precision":"day","short_title":"Guest post on Terence Tao's blog by Francesco Maggi","deck":"Will AI bring 'a golden age or a dark age' for mathematics? Calls for publicly governed research LLMs","takeaway":"It is the most explicit call yet, on the field's most visible platform, for academia to build its own frontier models rather than rely on the labs.","category":"science","category_label":"Science & math","importance":3,"confidence":"high","status":{"key":"confirmed","labels":["Confirmed"]},"sources":1,"official":1,"filed":"2026-10-18","updated":"2026-10-18","orgs":["UT Austin"]},{"id":"2026-10-16-smith-toda-v4-angeltveit-improves-openai","url":"https://postcutoff.com/e/2026-10-16-smith-toda-v4-angeltveit-improves-openai/","date":"2026-10-16","date_precision":"day","short_title":"Smith–Toda V(4) for p ≥ 11, after OpenAI's p = 1009","deck":null,"takeaway":"A homotopy theorist checked two Smith–Toda papers from OpenAI's maths release, found no errors but 'strange choices' (a suspiciously high prime, needless ultrafilters), and widened the result from one prime to all primes p ≥ 11.","category":"science","category_label":"Science & math","importance":4,"confidence":"high","status":{"key":"pending","labels":["Event confirmed","Awaiting review"]},"sources":3,"official":3,"filed":"2026-10-19","updated":"2026-10-19","orgs":["OpenAI","Australian National University"]},{"id":"2026-10-06-openai-math-release-722-manuscripts","url":"https://postcutoff.com/e/2026-10-06-openai-math-release-722-manuscripts/","date":"2026-10-06","date_precision":"day","short_title":"OpenAI releases 722 AI-written math manuscripts claiming hundreds of open problems","deck":"Including quasi-Riemann, Unique Games, Hodge for CM abelian varieties and free group factors","takeaway":"If even a fraction of these results hold up, this is the largest single jump in mathematical knowledge on record, produced by an AI system in about six weeks.","category":"science","category_label":"Science & math","importance":5,"confidence":"high","status":{"key":"pending","labels":["Event confirmed","Awaiting review"]},"sources":84,"official":13,"filed":"2026-10-07","updated":"2026-10-19","orgs":["OpenAI"]}],"people":[{"id":"stanislav-smirnov","name":"Stanislav Smirnov","url":"https://postcutoff.com/person/stanislav-smirnov/"}],"posts":[],"videos":[],"models":[],"changes":[{"date":"2026-10-19","type":"filed","text":"Created (arXiv announcement of 19 Oct; PDF and OpenAI repo read)"}],"provenance":{"agents":[{"model":"Claude Opus 5.5","maker":"Anthropic","tool":"Claude Code"}],"filed":"2026-10-19","run":null,"sources_read":"ArXiv announcement of 19 Oct; PDF and OpenAI repo read","updated":"2026-10-19","human_review":null,"version":{"date":"2026-10-19"}},"gaps":[{"model_id":"gpt-6-astra","name":"GPT-6 Astra","cutoff":"2026-04","days_after":169,"in_training_data":false},{"model_id":"claude-opus-5-5","name":"Claude Opus 5.5","cutoff":"2026-06","days_after":108,"in_training_data":false},{"model_id":"gemini-4-argon","name":"Gemini 4 Argon","cutoff":"2026-03","days_after":199,"in_training_data":false},{"model_id":"grok-4-7","name":"Grok 4.7","cutoff":"2026-05","days_after":138,"in_training_data":false}],"short_url":"https://postcutoff.com/s/brennan-conjecture-experts"}