AHM Statement on the Announced Proof of the Riemann Hypothesis

AHM communications group

Today, November 20, 2026, a mathematician at Anthropic announced the existence of an AI-generated Lean program confirming the Riemann Hypothesis, long a guiding open problem in mathematics.

This announcement is a major departure from traditional methods in mathematical research, representing a new value system put forward by AI corporations that foregrounds technical validity of logical statements over accountability, clarity, accessibility, and understanding. At this time, no human being can reasonably claim to be responsible for, or even to understand, such a document. Neither can any mathematician confirm with full certainty that the proof is correct. Worse, this corporate model of proof production seems designed to excuse humans from responsibility for scientific work. Instead, this model asks scientists to entrust their research to machines that AI corporations control.

Autonomous computer programs that investigate deep mathematical questions are now instruments of power in scientific research, and scientific power enables other types of power that can directly threaten human lives. This power is now concentrated in the hands of AI companies that seem increasingly untethered from the interests of human society. We believe that AI corporations urgently require democratic oversight. Milestones such as the Riemann Hypothesis are used by AI companies as little more than publicity for the superintelligence that they wish to build. We are gravely concerned that this uncontrolled growth poses a threat to society writ large.

Mathematicians have seen their world inverted by AI disruption in the last several months, and we have experienced firsthand how AI impacts human communities in practice. AI corporations hold an impoverished view of intelligence and human achievement. As a model for scientific research, the overproduction of unreadable mathematical documents pales in comparison to the richness of the mathematical dialogue that humans have conducted for millennia. What makes the Riemann Hypothesis valuable is its role in guiding the development of large areas of modern mathematics. Today’s announcement builds upon over a century of human work, but fails to advance mathematical understanding in the same way. Rather than accepting this model, we urge mathematicians to build a future where our research is free from dependence on AI models and their corporate owners.


Crossposted from the AHM website.


Received 20 November 2026.

53 responses to “AHM Statement on the Announced Proof of the Riemann Hypothesis”

  1. Aperson Avatar
    Aperson

    Could you please stop posting an announcement every time AI proves a big result? At least do something. Or is this all you can do? Asking people to stop using AI is pointless: if even one person uses it, your perfect world falls apart. Wake up and accept reality.

    1. anotherpersone Avatar
      anotherpersone

      Engaging in seminars, reading papers, teaching math, thinking about concepts, struggling with difficult proofs, drawing at the board. Yeah, I am pretty sure we are doing something. Are you?

    2. Pádraig Daly Avatar
      Pádraig Daly

      How is continuing to do actual mathematics doing nothing? Preserving the human culture, teaching, researching, understanding – this is not doing nothing.

      The 17 million lines of unintelligible computer code may eventually lead to something worthwhile but it will be through the activities of mathematicians in groups like this, who actually engage with mathematics.

      1. Nomen Avatar
        Nomen

        And I bet they are also breathing and walking and eating and drinking, this is not doing nothing.

        I am pretty sure some if not most of the group refuse to interact with mathematical artifacts produced by AI whatsoever, so no, this group in particular is not going to help organizing the new ideas in the AI proof.

    3. ashvinrajan Avatar
      ashvinrajan

      So it would equally meaningful for you if AI posted innumerable lines of unintelligible code saying the Riemann hypothesis (or for that matter all the problems solved) were all false? What do you think mathematics is about?

  2. coralcheerfully5084a986fd Avatar
    coralcheerfully5084a986fd

    “As a model for scientific research, the overproduction of unreadable mathematical documents pales in comparison to the richness of the mathematical dialogue that humans have conducted for millennia.” Strongly disagree with that. Many beautiful ideas have already come from AI generated math.

    “Rather than accepting this model, we urge mathematicians to build a future where our research is free from dependence on AI models and their corporate owners.” Do you have a realistic plan for that future? This sounds like saying, after cars were invented, “We won’t accept a world with cars because they are too fast and they can kill people. Let’s pretend they were never invented.”

    1. Anonymous Avatar
      Anonymous

      Why always the car analogy just because it supports your view?

      Radium in watches was banned, cigarettes are heavily discouraged, nuclear energy is banned in Germany, DDT is banned (extremely regulated), thalidomide is banned etc.

      All of these had superficially legitimate use cases initially, DDT is superb for eradicating malaria for example. I added nuclear energy to show that political will can ban technology, not to make a statement that it should be banned.

      1. coralcheerfully5084a986fd Avatar
        coralcheerfully5084a986fd

        Please refute my argument directly instead of focusing on a minor point in the analogy. I’m talking about the mathematicians in AHM who daydream about a “clean” future but offer no realistic plan for achieving it.

        They can choose not to smoke when others do, and they can choose to walk everywhere instead of driving. But if they issue a statement every time someone breaks a speed record in a car, repeating the same points without offering any concrete plan, it becomes ridiculous.

        1. Anonymous Avatar
          Anonymous

          Repeating things online is THE method of Silicon Valley to shape public perception and desires. Silicon Valley has had information dominance regarding AI for over 4 years.

          You could not go online for more than one minute without hearing how awesome AI is.

          Shouting down opposition, like it is done after every AHM statement, is part of the information dominance strategy.

          1. coralcheerfully5084a986fd Avatar
            coralcheerfully5084a986fd

            Does AHM have any plan other than writing statements, about how to shape the future of math? At least people in SV are doing something huh.

            1. someone Avatar
              someone

              yes, they plan to do math

      2. ashvinrajan Avatar
        ashvinrajan

        These superficial analogies are absurd.

    2. Nilima Nigam Avatar
      Nilima Nigam

      There are several reasonable plans for the kind of future the post envisions. Here’s one:

      https://proofsandprompts.com/2026/09/30/open-letter-to-the-european-mathematical-society/

      SAIR is another. https://www.openmathmodel.org/about/

      To be clear, the post does not advocate eschewing the use of AI in mathematics. It advocates eschewing the -model- they describe: generation of unreadable ‘Lean-formalized-ergo-correct’ mathematics on scale with no attempt at human readability, and an extreme reliance on AI tools run by big corporations.

      In your own analogy, if Ford had launched a car series in which automobiles spewed toxic waste on public thoroughfares, people should look for better purveyors of automobiles.

  3. Nomen Avatar
    Nomen

    Yes, we should go back to the good old days, when journal papers foregrounded accountability, clarity, accessibility, and understanding over technical validity of logical statements, when journal corporations held no scientific power and was under democratic oversight. After all, submitting papers to journals is the traditional method of mathematical research that humans have conducted for millennia, and most importantly, every paper published in a journal leads to immediate human understanding after a single day.

  4. Anonymus Avatar
    Anonymus

    Hi,

    IMO, to claim that the models are not advancing mathematical understanding, or that a proof of the Riemann Hypothesis does not advance mathematical understanding is a facially absurd claim. It is astounding to me that the AHM feels comfortable putting out a statement claiming this. It is of course true that nobody yet understands this proof of RH, or of many AI-proven results from the recent past. But equally so, it is true that having access to these unreadable slop documents, the huge number of new ideas that come from these resolutions to significant problems, etc, is of tremendous significance.

    One can lament the beautiful mathematics that might be lost due to AI, but clearly there is far, far more beautiful mathematics and wonderful ideas that have been found by the models, by humans collaborating with them. Asking mathematicians not to employ such capable and brilliant oracles is worse for the subject than the alternative. Clearly, this is a golden age for mathematics (though unfortunately, not a golden age for us mathematicians)

    On the non-mathematical points, regarding the risk posed by these models to upset our society, I cannot disagree. The way we are currently organized as a society is just not going to work in the near future.

    1. ashvinrajan Avatar
      ashvinrajan

      You make several valid points. However, AI is relying on millenia of work done by humans, and it is unclear whether this approach to mathematics will eventually advance the subject. Several young mathematicians are now abandoning possible carrers in the subject. Can AI advance mathematics all by itself? I do not believe that it can. This is how AI companies are using AI-proposing it ultimately as an alternative to human mathematics and benefitting from the reputation of the subject in one way or another.

      1. Nomen Avatar
        Nomen

        What exactly do you mean by “advance”? The AHM supporters seem to really like redefining words, but if you ask any mathematicians just 8 months ago whether a legit proof of RH counts as “advance”, there would be exactly zero person answering “no”.

  5. Alexander Avatar
    Alexander

    Does this “mathematician at Anthropic” – the person who will go down in history as having proved the Riemann Hypothesis – have a name?

    1. Pierre Menard, Author of the Quixote Avatar
      Pierre Menard, Author of the Quixote

      Yes, he is the Mozart of Slop

      1. Grigori Avramidi Avatar
        Grigori Avramidi

        Begs the question of who is his Da Ponte?

        1. Sofia Duiniúil Avatar
          Sofia Duiniúil

          Unfair comparison; considering the three Da Ponte operas as “complete works of art”, not in the Wagnerian sense but simply taking the music, text and dramaturgy all together conjunctively, Mozart was still the guy that did most of the work, and Da Ponte’s libretti on their own aren’t exactly milestones of literature — they are libretti as libretti ought to be. In fact I think the reason they are praised is that brilliant music was composed upon them, not the other way around, that they and they alone made such music possible.

          1. Grigori Avramidi Avatar
            Grigori Avramidi

            I didn’t make a comparison. I asked ”who is his Da Ponte” precisely because I couldn’t think of a comparison and it was interesting, to me if someone could think of one.

            And, you can simulatenously think that Mozart is a genuis (obviously) and that structurally Cosi, Marriage, and Don Giovanni really are more interesting and light-footed than, say, Magic flute or Idomeneo.

    2. Pádraig Daly Avatar
      Pádraig Daly

      Claude I believe

    3. Anonymous Avatar
      Anonymous

      Herostratus perhaps?

    4. Victor Avatar
      Victor

      This people fail at the most basic credit attribution for the most important result of the century. Could it be that all this talk about ethics is really about ego?

  6. Nomen Avatar
    Nomen

    What exactly is the role of RH itself (as opposed to the broader study of zeta/L functions) “in guiding the development of large areas of modern mathematics”? Attempts to prove FLT gave rise to the modern algebraic number theory, there doesn’t seem to be a similar story for RH. Unlike FLT, what makes RH valuable is its numerous consequences, indeed, this would be the answer from most mathematicians if one asked “why is RH important” a year ago.

  7. Ano Avatar
    Ano

    In my view, Levent hasn’t hyped anything this time. It’s AHM that’s using this opportunity to push their values.

    1. ashvinrajan Avatar
      ashvinrajan

      He doesn’t need to.

    2. elegantc7df67a5d5 Avatar
      elegantc7df67a5d5

      Oh no. He just published on the day that the Anthropic IPO failed.

      Maybe we get a proof of the Birch and Swinnerton-Dyer conjecture when (not if) ANT hits $50.

      1. Nomen Avatar
        Nomen

        The funny thing about “unbridled capitalist kleptocracy” is that you can bet against it on the market. Put your money where your month is.

        1. Anonymous Avatar
          Anonymous

          Timing the bets against nepotism and kleptocracy is notoriously hard, since you don’t know when exactly companies like Nvidia are allowed to fail.

          .Doesn’t Anthropic have an entrance exam for its influencers?

          1. Nomen Avatar
            Nomen

            Yeah, it makes total sense for an Anthropic influencer to urge people who are saying its stock is going to crash, that they should actually short the stock.

  8. Ano Avatar
    Ano

    In my view, Levent hasn’t hyped anything this time. It’s AHM that’s using this opportunity to hype their values.

  9. Prahlad Avatar
    Prahlad

    I would urge all mathematicians to simply stop engaging with these companies.

    Do not give them your time. Do not give them your money. Do not give them credit. Use AI tools, by all means, but do not let them dictate the public discourse.

    Everytime a mathematician tells me that they want to run a seminar to understand a particular result from OpenAI’s slop drop, I simply don’t understand why they are doing it.

    When India fought for its independence from an exploitative government, it used civil disobedience as a weapon. I call on everyone who reads this to follow that lead, and simply disengage from this never-ending advertisement.

    1. Nomen Avatar
      Nomen

      Yeah, why do these mathematicians want to understand a proof? Maths is about our collective fight against capitalism, corporatism, fascism, etc., etc., not about some mathematical proofs!

    2. Sirawit Pongnakintr Avatar
      Sirawit Pongnakintr

      I would urge all mathematicians to simply stop engaging with AHM.

      Don’t let AHM say what is mathematics and what is not. Don’t let the AHM speech “Mathematicians did not ask for this work to be done” stop you from working on what you like to do.

      You have your rights to use or not use AI. You have your rights to care or not care about AI-produced results.

      You do not have your rights to stop other people from developing AI, from using AI, or from studying mathematics coming from AI. You can advocate as much as you want, but physically removing my access to AI is not going to happen soon, and I’m going to stand my ground and fight for my rights.

      We must know, we will know, and we will not stop there. We will also understand things, regardless of coming from humans or AI.

  10. Giovanni Mongardi Avatar
    Giovanni Mongardi

    This is not the first time that someone claims to have proven Riemann’s Hypothesis. However, we should consider it proven only when a mathematician, or a group of mathematicians, can confirm its correctedness.

    1. ashvinrajan Avatar
      ashvinrajan

      To wit, the humans have struck back, Alessandro Masullo has been working on the GRH for months and Ravi Vakil decided to publish his work early because of this AI drop, The approaches claimed by Anthropic and Masullo are different and will eventually contribute to understanding and new mathematics. But again, without humans to understand, interpret, and contextualize all this work it will just be slop. I also believe that these efforts by humans “the richness of the mathematical dialogue that humans have conducted for millennia” are essential to the creation of new mathematics.

      1. Anonymous Avatar
        Anonymous

        Masullo’s work is incorrect and has massive gaps. It is one more in the line of litany of incorrect proofs of the Riemann Hypothesis. Humanity did not strike back, humanity is just using it to cope. The faster we stop coping, the faster we can start adapting.

  11. Fred Daum Avatar
    Fred Daum

    The AHM should actually read Levent’s honest and inspiring 4 page remark on his work with Claude before reacting so negatively. This is a beautiful day for humanity, which invented the RH, invented math, invented computers, invented AI, and invented all the deep math theories used to prove the RH. More generally, applied mathematicians, physicists, chemists, biologists, engineers and most pure mathematicians are appalled at the infantile rants of the AHM like this one.

    1. Pierre Menard, Author of the Quixote Avatar
      Pierre Menard, Author of the Quixote

      “Inspiring” LMAO

      1. elegantc7df67a5d5 Avatar
        elegantc7df67a5d5

        “honest and inspiring” is Claude language. Almost “load-bearing”:

    2. elegantc7df67a5d5 Avatar
      elegantc7df67a5d5

      You left out the IP theft, stock market bubbles, IPO fraud and concentration of capital in your childlike fanboy comment.

      1. Nomen Avatar
        Nomen

        Defeat le evil corporations by defending IP laws1!1!

        1. Anonymous Avatar
          Anonymous

          Yes. IP laws have always protected individuals, that is why artists were in favor of them even if corporations took a cut.

          Silicon Valley’s web economy has always run on theft, so their propaganda has been to abolish IP laws and become the new middlemen. Resurfacing Slashdot sentiments from 2000 when the enemy was Microsoft or Disney is an anachronism.

          1. Nomen Avatar
            Nomen

            Unless you want to argue that IP is somehow a fundamental part of human nature ubiquitously, otherwise the “IP theft” claim is purely legal, so just sue them and see the result.

    3. Nilima Nigam Avatar
      Nilima Nigam

      It’s wise to deploy quantifiers while making strong assertions about the views of groups of people. It is not universally true that ‘applied mathematicians, physicists, chemists, biologists, engineers and most pure mathematicians are appalled at the infantile rants of the AHM like this one.’

      I’m an ‘applied mathematician’, and actually agree with this post of the AHM.

      You’re of course free not to agree with the post. Some others may share your view. And some may disagree.

  12. Scott Guthery Avatar
    Scott Guthery
    1. Harvard and MIT are both corporations that very much look after their own interests and have very active public relations departments.
    2. I have to wonder how many members of the AHM understand and can explain Wiles’ proof of Fermat’s Last Theorem.
    1. elegantc7df67a5d5 Avatar
      elegantc7df67a5d5

      Harvard and MIT didn’t steal all human knowledge and had to settle for $1.5 billion for criminal torrenting like Anthropic did.

      What is the argument here anyway? Harvard and MIT are suboptimal, so we can as well allow an unbridled capitalist kleptocracy?

      1. Nomen Avatar
        Nomen

        Have your knowledge be stolen by AI companies? Because it does appear that you no longer possess it.

        1. Anonymous Avatar
          Anonymous

          Thank you for another recycled Slashdot comment.

  13. Anatol Wegner Avatar
    Anatol Wegner

    “Autonomous computer programs” ? – why go with the AI labs discourse here? if anything Alpoge’s RH and Buckmaster’s unforced NS show that these results required extensive expert guidance by humans. See Buckmaster’s AI statements:

    “We used Anthropic models (Claude) throughout this work, and no others. The work builds on unreleased work of the authors on computer-assisted proofs, in particular concerning the non-uniqueness of Leray–Hopf solutions in the style of Jia and Šverák [65]
    and blow-up for fluid equations with boundary. It builds also on numerical work of the authors on finding self-similar profiles, on our earlier papers on singularity formation, and on our unreleased drafts on the forced hypodissipative Navier–Stokes equations and the forced Navier–Stokes equations. The idea of pursuing discretely self-similar solutions came from the paper of the second author with Coiculescu and Li on non-uniqueness [17]. The agents were given a curated dossier of potential leads and then guidance after our strictly-uncontaminated experiment producing unforced Euler and forced Navier–Stokes blow-up from our forced Euler and forced hypodissipative Navier–Stokes manuscripts. A simpler mechanism sufficed to produce unforced hyperdissipative Navier–Stokes blow-up, i.e. for ν ·(−∆) and then a heavier computation produced this profile. We used the enslopification skills of Scott Armstrong [10] to create a draft from an initial Lean proof (Subsection 1.5); the draft followed that proof step by step. We then rewrote it with Claude under our direction, citing standard results in place of proofs and simplifying the argument over successive revisions, so that the proof given here no longer follows the Lean proof.”

  14. Daniel Hadas Avatar

    Such statements carry no weight unless endorsed by a list of respectable people. The AHM has a list of ~1000 members, but apparently there is no public evidence that their members have approved of this statement.

Comments are moderated. Read our comment policy.

Add to the discussion

New posts by email.

Prefer a feed reader? Subscribe by RSS.

Also on Mathstodon.

Latest comments across the site.