Will CMI declare the Riemann Hypothesis solved by December 31, 2027?

- [2026-09-11] The official Clay Mathematics Institute Millennium Problems page (last modified this date) currently lists the Riemann Hypothesis under "Unsolved problems," while listing Navier-Stokes separately under "Active problems" and only the Poincaré Conjecture under "Solved problems" — confirming RH has no qualifying CMI declaration as of the fetch date. (https://www.claymath.org/millennium-problems/) - [2026-09-08] OpenAI announced a claimed AI-generated solution to a version of the Navier-Stokes Millennium Problem (not Riemann Hypothesis); "OpenAI announced that a multi-agent AI system had produced 'an analytical proof and a Lean formalization that an initially smooth fluid at rest can develop a singularity in a finite time' for the 3D Navier-Stokes equations, one of the Clay Mathematics Institute's seven Millennium Prize Problems." This is unrelated to RH but shows the current climate of AI-driven Millennium Problem claims. (https://www.tuftsdaily.com/article/2026/09/mathematicians-still-checking-the-navier-stokes-proof-that-openai-claims-to-have-solved) - [2026-09-21] Even the Navier-Stokes claim remains contested/unverified weeks later, with Scientific American noting "OpenAI's solution exploits an external force to break the Navier-Stokes equations, which are supposed to describe how fluids, such as water, flow" — illustrating that CMI has not moved to certify even the more advanced current AI claim, let alone one for RH. (https://www.scientificamerican.com/article/did-openai-solve-the-wrong-navier-stokes-problem/) - [2026-09-11] Wikipedia's Clay Mathematics Institute article corroborates that Navier–Stokes progress "OpenAI published a solution in September 2026 however the result has yet to be verified by the Clay Institute or the independent mathematical community... and is the subject of a priority dispute" , with Riemann Hypothesis still listed as unsolved. (https://en.wikipedia.org/wiki/Clay_Mathematics_Institute) - [2026-08-10] Anthropic published a research note (updated 2026-08-13) stating "An unreleased Claude model improved the lower bound for the fraction of zeroes of the Riemann zeta function that satisfy the hypothesis, raising it from 41.6% to 67.2%" — a partial technical advance, explicitly not a full proof. (https://www.anthropic.com/research/riemann-zeta) - [2026-08-10] Coverage of the Anthropic result was explicit that this did not solve RH: "Claude didn't prove the hypothesis. Nobody expected that." (https://www.datacamp.com/tutorial/claude-and-the-riemann-hypothesis) - [2026-09-09] A third-party tracker summary states "As of September 9, 2026, the Riemann Hypothesis remains unsolved, lacking any accepted proof or disproof within the mathematical community" . (https://www.octagonai.co/markets/science-and-technology/physics-math/when-will-the-riemann-hypothesis-be-solved) - [2026-09-05/09-01] The same tracker reports market volatility driven by misinformation: "In August 2026, Anthropic's Claude AI model produced a significant mathematical result related to the Riemann zeta function, which was incorrectly reported as a solution to the Riemann hypothesis... This misinformation, despite Anthropic's explicit clarifications, caused confusion in media and on social media platforms" , followed by a correction. (https://www.octagonai.co/markets/science-and-technology/physics-math/when-will-the-riemann-hypothesis-be-solved) - [2026-09-09] A survey of AI-assisted math activity notes on RH specifically: "An unreleased research version of Anthropic's Claude improved the long-standing lower bound on the fraction of non-trivial zeta zeros lying on the critical line from 41.6% to 67.2%... Mathematicians Levent Alpöge and Ralph Furman at Anthropic validated the result but found it genuinely hard to follow" — still framed as partial progress, not a solution. (https://securemachinery.com/2026/09/09/millennial-math-problems-lead-to-a-surge-in-ai-assisted-activity/) - [2026-04-15] CMI/Harvard CMSA held a dedicated Millennium Prize Problems lecture on the Riemann Hypothesis by Peter Sarnak, part of a series treating all remaining problems (except Poincaré) as open; the series description states "In 2000, the Clay Mathematics Institute identified seven significant open problems, and offered a $1 million bounty for each. Of these, only the Poincaré Conjecture has been resolved." (https://cmsa.fas.harvard.edu/millennium) - [2026-09 (ongoing)] Numerous non-CMI, non-peer-reviewed "proof" postings (e.g., independent researchers, preprint sites) continue to circulate claiming RH is solved, but none have been accepted by CMI or the mathematical community; the resolution criteria explicitly exclude such postings. - [2026-09-26] No CMI news release, statement, or updated problem-page status change announcing a Riemann Hypothesis solution has appeared; the CMI news page and millennium-problems page show no such update as of the last fetch. NOT FOUND: No official CMI statement, news release, or press communication referencing a solved or disproved Riemann Hypothesis; no evidence of any Millennium Prize being awarded or scheduled for RH; no confirmed off-critical-line zero or accepted full proof/disproof by the mathematical community. BOTTOM LINE: As of late September 2026, CMI's own website explicitly lists the Riemann Hypothesis as unsolved, and recent developments (Anthropic's partial zero-density bound improvement, OpenAI's separate and still-contested Navier-Stokes claim) show only incremental or unrelated progress — nothing indicating CMI is near declaring RH solved before the market's late-2027 deadline.

- [2000-05-24] The Clay Mathematics Institute established seven Millennium Prize Problems, each carrying a $1 million prize, at a Paris announcement; RH was one of the seven original problems. https://www.claymath.org/millennium-problems/ - [2026-09-11] Per CMI's own current problem list, only the Poincaré Conjecture is marked "Solved"; the Riemann Hypothesis remains listed under "Unsolved problems," alongside Hodge, P vs NP, BSD, and Yang-Mills. https://www.claymath.org/millennium-problems/ - [2026 (page updated)] Reference sources confirm to date, only one of the Millennium Prize Problems, the Poincaré conjecture, has been solved , giving an outside-view base rate of roughly 1-in-7 (14%) of Millennium problems ever resolving "Yes" over the ~26 years since 2000, and 0-for-6 for the other six including RH. - [2003→2010] The one historical success took years even after a proof existed: a correctly solved Millennium Problem must be submitted to a CMI-approved journal, pass peer review, and then withstand a two-year waiting period before the prize is formally awarded — this process is deliberately slow, and Perelman's case showed why ; his proof was submitted informally on arXiv between 2002 and 2003 and required multiple independent verification teams working over several years before consensus emerged . Official CMI declaration only came in 2010, ~7 years after the proof first appeared. - [2026] Current expert/quant surveys of the specific RH timeline: prediction markets do not assign a high probability to a near-term solution; Metaculus forecasters, for example, estimate a median resolution date of April 2044 . - [2026-04] Mathematician Peter Sarnak expressed an expectation in April 2026 that he would live to see the resolution of at least one major Millennium Prize problem, including the Riemann Hypothesis — a long-horizon, non-imminent framing from a leading expert in the field. - [2026-09-09] A Kalshi-linked market ("When will the Riemann hypothesis be solved?") priced "Before 2028" at only 20% (market) vs 11.7% (model estimate), and "Before 2030" at 44% (market) vs 27.7% (model) — both explicitly noting these estimates reflect post-hype correction, not genuine expectation of imminent proof. https://www.octagonai.co/markets/science-and-technology/physics-math/when-will-the-riemann-hypothesis-be-solved - [2026-08-09→09] An AI (Anthropic's Claude) improved a lower bound on zeros lying on the critical line from 41.6% to ~67.2%, triggering a large but short-lived speculative price spike; experts (Terence Tao among them) and Scientific American publicly clarified this did not constitute a solution to the Riemann hypothesis , and the market subsequently retraced most of the spike. https://www.octagonai.co/markets/science-and-technology/physics-math/when-will-the-riemann-hypothesis-be-solved - [2013-2026] Numerous informal/arXiv "proof" claims of RH have surfaced over the decades (e.g., 2013, 2014, and multiple 2026 preprints/working papers such as a May 2026 Cambridge Open Engage submission and a January 2026 ResearchGate paper), none of which have been accepted by the mathematical community or CMI — illustrating a long track record of false-alarm claims with zero historical conversion into an official CMI declaration. https://www.researchgate.net/publication/400157251_The_Proof_of_the_Riemann_Hypothesis - [Base rate synthesis] Across 26+ years since the prize was created, across all seven problems, only one "Yes" resolution (Poincaré) has occurred, and that resolution took ~6-10 years from informal proof posting to official CMI declaration — no comparable "fast" resolution exists in this reference class, and the process (peer review + mandated 2-year wait) is structurally slow even after a valid proof appears. https://www.claymath.org/millennium-problems/ BOTTOM LINE: The reference class (Millennium Prize problems) has produced exactly one CMI "solved" declaration in 26 years, and that single case took roughly 6-10 years from proof to official declaration due to CMI's peer-review-plus-two-year-wait rule; current expert and quantitative forecasts (Metaculus median ~2044, prediction-market odds of ~12-20% even by 2028-2030 for RH merely being "solved" mathematically, let alone officially declared by CMI) all point to a very low base rate for a CMI declaration on RH specifically by end-2027. NOT FOUND: No historical instance of CMI ever issuing a "disproof/counterexample" declaration; no public statement from CMI itself estimating a timeline for RH; no evidence of any current candidate proof that has entered CMI's formal review pipeline.

- [2026-09-09] An independent, peer-reviewable paper by Youness Lamzouri gave a "new, conceptually simpler, unconditional proof" reproducing and slightly extending the Anthropic/Claude result, explicitly stating "A proof of these results was very recently produced by an internal research version of Claude developed by Anthropic and subsequently verified by Alpöge and Furman... This argument is technically intricate, and its main mechanism is not immediately transparent" — showing the AI-derived advance is durable enough that a human number theorist could confirm and generalize it within one month, evidence of an unusually fast human-AI verification loop. (https://arxiv.org/pdf/2609.02882) - [2026-09-09] The same paper's appendix reports that "AxiomProver, an AI system currently under development by Axiom Math, autonomously generated (from natural language) an unconditional formal certificate for Proposition 2.1," with the Lean certificates posted publicly and produced "on very short notice" after the proof's completion — a concrete, dated demonstration that formal (Lean-verifiable) certification of frontier zeta-zero results can now be generated in days rather than years. (https://arxiv.org/pdf/2609.02882) - [2026-09-09] This shows the mechanized-proof-checking infrastructure (Lean + AI provers) that could, in principle, shortcut CMI's traditionally slow, multi-year human-review bottleneck if a full RH proof were ever produced, since partial results are already being auto-formalized essentially in real time. (https://arxiv.org/pdf/2609.02882) - [2026-09-09] Despite the caveats, the pace of improvement on RH-adjacent bounds is compounding quickly: the critical-line zero-proportion bound moved from 41.6%→67.2% (Anthropic, Aug 2026) and was then reproduced/extended to sharper related constants (67.25% simple-on-critical-line, 83.62% distinct, 88.76% simple-or-critical) within roughly one month by a human-authored, unconditional proof. (https://arxiv.org/pdf/2609.02882) - [2026-09-09] A Kalshi-linked market update explicitly frames the state of play as "ongoing research explores new methods" with ongoing 2026 work "further" pushing verified unconditional bounds, i.e., market makers acknowledge a live research frontier rather than a stalled one, even while pricing "Before 2028" at only 20% (market) vs 11.7% (model). (https://www.octagonai.co/markets/science-and-technology/physics-math/when-will-the-riemann-hypothesis-be-solved) - [2026-09 undated] Independently, some thinly-traded prediction-market users have priced meaningfully higher odds of an AI-linked resolution within the relevant window, e.g. one Manifold market shows a forecaster assigning "66% chance" to "OpenAI discovers the first proof of the Riemann Hypothesis before 2027" and another shows "72% chance" that "a Human or AI model" will solve RH "by 2028" — illustrating that a non-trivial minority of informed bettors do not rule out a resolution landing inside the question's window, even though these are low-liquidity, non-expert-consensus markets. (https://manifold.markets/HarrisonLucas/ai-program-resolves-riemann-hypothe) - [2026-09-08] The same week saw OpenAI's claimed AI-generated solution to the Navier-Stokes Millennium Problem, establishing a fresh, live precedent this year for a major AI lab publicly asserting a Millennium Prize-level result and pushing it into community/CMI-adjacent scrutiny — a template (AI claim → rapid arXiv counter-verification → Lean formalization) that could be replicated for RH if a genuine full proof emerged, compressing the "informal proof to community consensus" phase relative to the historical Poincaré timeline. (https://en.wikipedia.org/wiki/Clay_Mathematics_Institute) - [2026-09-23] CMI's own 2026 Clay Research Conference (Oxford, 23 Sept) continued to actively convene top analytic number theorists working adjacent to zeta-zero statistics, keeping CMI directly and currently engaged with the specific research community from which any qualifying RH proof would have to originate, rather than being a passive, disengaged institution. (https://www.claymath.org/) - [2026] The resolution criteria for this market itself lower the bar relative to historical Poincaré-style review: they state a qualifying declaration can be "a positive proof, counterexample, disproof, or any other resolution accepted by CMI," and explicitly that AI involvement or a recipient declining the prize "will not be considered" as disqualifying — meaning CMI's rules do not require a human-only, decades-long process, and an AI-originated, rapidly Lean-verified proof (as the Sept 2026 partial-result pipeline demonstrates is now technically feasible) would not be barred from qualifying if completeness were ever achieved. BOTTOM LINE: The single most plausible (though still low-probability) path to a YES resolution is an acceleration of the AI-plus-formal-verification pipeline already visible in September 2026 — where Anthropic's Claude produced a partial result, an independent mathematician reproduced and extended it within weeks, and an AI system (AxiomProver) auto-generated Lean certificates almost immediately — compressing the traditional multi-year proof-to-CMI-declaration timeline enough that, if a genuinely complete proof or disproof surfaced via this fast-moving ecosystem before late 2027, CMI could plausibly move faster than the ~6-10 year Poincaré precedent suggests. NOT FOUND: No evidence of any full, community-accepted proof or disproof of RH itself (only partial zero-density bounds); no CMI statement indicating an expedited or AI-specific review track; no confirmation that CMI's traditional two-year waiting period would be waived even under this faster-verification scenario.

- **[2018-09-24, historical precedent]** The last time a top-tier mathematician (Michael Atiyah, then at Edinburgh) publicly claimed an RH proof, at the Heidelberg Laureate Forum, experts were skeptical because "many experts doubt its validity, especially because Atiyah, 89, has been making mistakes in recent years" , and the underlying paper his claim rested on had not even been published yet — the claim died quietly with zero CMI action, illustrating that celebrity/media attention alone has never once moved CMI on RH. - **[2018-09-26 CMI rule, still in force]** The official CMI rules page (last modified 2024-06-17, currently live) states three conjunctive gating conditions before CMI will even *consider* a solution: Before CMI will consider a proposed solution, all three of the following conditions must be satisfied: (i) the proposed solution must be published in a Qualifying Outlet (see §6), and (ii) at least two years must have passed since publication, and (iii) the proposed solution must have received general acceptance in the global mathematics community . This means even a hypothetical flawless RH proof submitted today (Sept 2026) could not clear the two-year clock before roughly September 2028 at the earliest — after this market's December 31, 2027 deadline — unless it had already been published and clocked two years by an unrealistically early date. - **[undated, mechanism]** The rules explicitly bar shortcuts: "CMI does not accept direct submission of proposed solutions" , meaning no lab or mathematician can simply email CMI a proof to start the clock; it must first clear a refereed "Qualifying Outlet" and then survive the community-acceptance test, adding friction beyond the two-year wait itself. - **[2010, only precedent]** In the sole historical "Yes" resolution, the institute's rules require that it waits two years after the result has been published in a refereed journal before awarding the prize, in order to give the mathematical community time to find any errors , and even then Perelman's proof — first announced in 2002 with a sketch published online, required dozens of mathematicians laboring for several years to fill in details and verify validity — took roughly seven-plus years total from first posting (2002-03) to CMI's award (2010), despite never even undergoing the formal journal-refereeing route. - **[2026-09-09, current independent-market pricing]** A live Kalshi/Octagon tracker gives RH being solved (mathematically, not even CMI-declared) "Before 2028" at only "Our model puts Before 2028 at 12% (market 20%)" and full solution by 2032 at just a 25% chance of solution by 2032... As of August 2026, prediction markets on Kalshi indicate a 25% implied probability that the Riemann Hypothesis will be solved by 2032 — these are odds on a bare mathematical solution existing at all, a necessary but not sufficient precondition for the much harder CMI-declaration bar this market requires. - **[2026-09-20, direct market read]** Polymarket's own live book for this exact question currently prices "Yes" for the December 31, 2027 date at just "Will CMI declare the Riemann Hypothesis solved by ___? ... 16% December 31, 2027" , i.e. the crowd itself assigns roughly 5-in-6 odds against. - **[2026-09, sister-market corroboration]** A related Polymarket contract on which Millennium Problem AI solves next shows "sub-1% probabilities on P versus NP, Riemann Hypothesis, and Yang-Mills indicate limited current AI focus or demonstrated capability on those fronts" , while attention and capital are concentrated instead on Hodge and Navier-Stokes — RH is explicitly the AI labs' lowest-priority target, not next-in-line. - **[2026-09-17, no CMI response even to bigger claim]** Even for the more advanced, more publicized Navier-Stokes counterexample claim, a Polymarket description notes that "Key near-term catalysts include independent mathematical scrutiny of the Navier-Stokes proof, public release of stronger models, and any confirmation or rebuttal from the Clay Mathematics Institute, all of which could accelerate or stall further announcements" — meaning CMI has not yet even weighed in on the more advanced case, let alone RH. - **[2026-09, base division of odds]** Bookmakers separately quote "AI lab announces another Millennium Prize solution by Dec 31 2027" (a much weaker bar than CMI declaration, requiring only a lab announcement) at 85%, and "OpenAI announces another Millennium Prize solution" at 81% — but these are lab-announcement markets explicitly distinct from and easier to trigger than a CMI declaration, and per "A qualifying announcement must express that the problem has been solved; announcements of partial results or progress toward a solution will not qualify" for those sister markets, underscoring how much rarer a full "solved" claim (let alone CMI ratification) is versus mere progress. - **[2026-03-05, expert survey]** A 2026 technical status review states plainly that "In 2026, after 167 years, the Riemann Hypothesis remains open," with only incremental numerical/asymptotic advances (twenty trillion zeros verified, de Bruijn–Newman constant pinned at zero) rather than anything resembling a candidate full proof entering any review pipeline. - **[ongoing]** Fringe "proof" postings continue appearing on preprint/blog platforms (e.g., a November 2025 Cambridge Open Engage working paper claiming to "prove the Riemann hypothesis by a new analytic approach," and independent-researcher sites), none peer-reviewed, none CMI-engaged, exactly the pattern the CMI rules are designed to filter out via the Qualifying-Outlet-plus-two-year-wait requirement. BOTTOM LINE: Both the structural rule (mandatory Qualifying-Outlet publication + a strict two-year community-acceptance clock that CMI has never waived) and the live market pricing (Polymarket's own RH-by-2027 contract at 16%, sister AI-focus markets pricing RH as the least-likely of the six remaining problems) point strongly toward NO; no candidate proof has even entered CMI's pipeline, and the sole historical "Yes" precedent took over seven years from initial claim to declaration. NOT FOUND: No evidence of any RH proof currently submitted to or under evaluation by CMI's Scientific Advisory Board; no CMI statement referencing a Qualifying-Outlet publication of an RH proof; no indication CMI has ever shortened or waived the two-year rule for any problem.