OpenAI Claims Three AI-Driven Scientific Breakthroughs; Independent Review Underway

Around October 27, OpenAI published an update on its collaborative research with the academic community, headlined by a striking claim: an internal model produced an alleged refutation of the Exponential Time Hypothesis (ETH) in computational complexity theory — namely, that 3SAT requires exponential time. If valid, it would overturn a number of established results that presuppose ETH. A circulating screenshot (relayed by @felpix) further claims the proof has since been formally verified in the Lean proof assistant.

Confirmed

• OpenAI officially announced the research progress, stating that model-assisted research refuted ETH (@EnvironmentalMix22).
• Theoretical computer scientist Ryan Williams (rrwilliams) announced that he, along with Russell Impagliazzo and Mohan Paturi, has been invited to organize a public, independent review of the refutation claim.
• Williams noted that the review team has signed no NDAs with OpenAI and receives no compensation, and pledged to keep the community updated with findings throughout the review; the team has formally requested that OpenAI promptly release all materials related to the ETH refutation claim.
• Fields Medalist Tim Gowers shared the news, noting that ETH is a major claim in complexity theory and will undergo independent review.

Not Yet Confirmed

• Whether ETH has actually been refuted: the review is not complete, and the verdict rests on its outcome.
• The claim that "the proof has been formally verified in Lean" comes only from a circulating screenshot and has not been confirmed by OpenAI or the review team.

Why It Matters

ETH is a long-standing core conjecture about the computational complexity of Boolean satisfiability (SAT), closely tied to P vs NP, and a large body of complexity results assumes it holds. A confirmed refutation would be a landmark achievement for AI-assisted mathematical research and would force the field to re-examine a host of theorems that rely on ETH. Meanwhile, the review team's arrangement — no NDAs, no compensation, full transparency throughout — offers a replicable model for how major scientific claims produced by AI can be subjected to human peer review.

2026-10-27 ~ 2026-10-27 · 45 related posts

Full story(3 episodes)→

• Episode 1: OpenAI Claims Three AI-Driven Scientific Breakthroughs; Independent Review Underway(2026-10-27, 45 posts)
• Episode 2: OpenAI Model Claims Refutation of ETH, Prompting Independent Review(2026-10-27, 2 posts)
• Episode 3: Claims of subexponential 3SAT algorithm stir P=NP debate(2026-10-27, 3 posts)

• OpenAI internal model claims to refute ETH; top theorists to run open verification — rrwilliams · 2026-10-27
• ETH verification team: no NDAs, no pay, raw data release requested from OpenAI — rrwilliams · 2026-10-27
• OpenAI models produce Lean-verified existence proofs for QCD and Yang–Mills — danintheory · 2026-10-27
• Harmonic announces three discoveries: ETH false, QCD existence, Kakeya conjecture — apples_jimmy · 2026-10-27
• [source] OpenAI PDF claims refutation of ETH, solving 3SAT in exp(n/(log n)^(1/2+o(1))) — rrwilliams · 2026-10-27
• OpenAI Announces Three New Discoveries in Math, CS, and Physics — thesaraharminta · 2026-10-27
• [source] Greg Brockman announces AI-assisted discoveries in math, CS, and quantum field theory — gdb · 2026-10-27
• 'The exponential time hypothesis is FALSE!' — TCS community reacts to 3SAT result — danintheory · 2026-10-27
• Bubeck hails subexponential-time 3SAT algorithm as biggest TCS surprise in 20 years — danintheory · 2026-10-27
• [source] OpenAI says its models produced a Yang-Mills existence proof, formalized in Lean — exophades · 2026-10-27
• Harmonic announces three discoveries in math, CS, and physics — thegautamkamath · 2026-10-27
• OpenAI claims its AI invented a new math technique, cracking a 35-year-old wave-focusing problem — 141_1337 · 2026-10-27
• Polymarket: OpenAI says its AI models made 3 major scientific discoveries in math, CS, and physics — Polymarket · 2026-10-27
• OpenAI claims its AI cracked decades-old problems in math, complexity theory and physics — THE_ILL_SAGE · 2026-10-27
• OpenAI announces AI-generated proofs targeting ETH, Kakeya, and Yang–Mills in one day — ChrisGPT · 2026-10-27
• Polymarket puts 54% odds OpenAI solves another Millennium Prize problem after 3 claimed AI discoveries — Polymarket · 2026-10-27
• OpenAI internal model claims to refute Exponential Time Hypothesis; independent review organized — Singularitarian · 2026-10-27
• AI announces new discoveries in harmonic analysis, complexity theory and QFT — MoonL88537 · 2026-10-27
• OpenAI claims models helped falsify the Exponential Time Hypothesis — Environmental_Mix22 · 2026-10-27
• Screenshot claims AI system refuted Exponential Time Hypothesis, verified in Lean — felpix_ · 2026-10-27
• "The Exponential Time Hypothesis is false": major complexity claim spreads — davidad · 2026-10-27
• OpenAI system disproves Exponential Time Hypothesis, result verified in Lean — FlorianGallwitz · 2026-10-27
• OpenAI announces three major breakthroughs, resurfacing a 2024 'P vs NP by 2027' bet — inductionheads · 2026-10-27
• Greg Brockman Announces Three Major AI-Assisted Scientific Discoveries — brhydon · 2026-10-27
• Viral 'OpenAI' paper also claims Kakeya Conjecture true in all dimensions — RexDouglass · 2026-10-27
• Paper attributed to OpenAI claims to refute the Exponential Time Hypothesis in 64 pages — wavefnx · 2026-10-27
• AI reportedly disproves the Exponential Time Hypothesis, shifting P vs NP odds — esrtweet · 2026-10-27
• ETH originators to lead open independent review of OpenAI's refutation claim — danintheory · 2026-10-27
• Claim: a Lean formalization 'proves' the Exponential Time Hypothesis is false — akbirthko · 2026-10-27
• OpenAI touts three major scientific discoveries in physics, math, and CS — pickover · 2026-10-27
• OpenAI claims Exponential Time Hypothesis is false, solving 3SAT in subexponential time — fortnow · 2026-10-27
• OpenAI claims its model refuted the Exponential Time Hypothesis, verified in Lean — AIFlow_ML · 2026-10-27
• Claimed disproof of Exponential Time Hypothesis: subexponential 3SAT algorithm announced — cephaloform · 2026-10-27
• Anthropic Co-founder Announces Three AI-Assisted Discoveries, Including Falsified ETH — inductionheads · 2026-10-27
• OpenAI claims subexponential 3SAT bound; Impagliazzo, Paturi, Ryan Williams lead open review — AIFlow_ML · 2026-10-27
• ETH proposers lead open review of solver claim; exponent shrink factor ~0.67 at a billion variables — AIFlow_ML · 2026-10-27
• If ETH falls, which conditional lower bounds survive on SETH? Napkin math inside — AIFlow_ML · 2026-10-27
• OpenAI announces three AI-made discoveries: ETH refuted, Kakeya conjecture, QCD existence — burny_tech · 2026-10-27

7 near-duplicate retellings: morey56 · rohanpaul_ai · Dr_Singularity · Dr_Singularity · Dr_Singularity · ErnestRyu · RileyRalmuto