Google DeepMind's AlphaProof Nexus solves dozens of open math problems

By Splitfeed EditorialLast updated Oct 12, 2026, 09:40 UTC

On October 12, 2026, Google DeepMind published results in the journal Science showing that its AlphaProof Nexus system had solved dozens of open mathematics problems, among them nine problems from the late mathematician Paul Erdős. Two of those nine had gone unsolved for 56 years. The publication marked a formal milestone in that the findings passed peer review before appearing, which the Science study itself notes as a distinguishing feature of how DeepMind chose to release the work.

The timing matters because it lands in the middle of a visible dispute over research norms in AI mathematics. OpenAI is currently facing a boycott from researchers over its practice of releasing AI-generated proofs directly to GitHub without peer review. DeepMind's decision to go through formal review before publication is therefore not just a procedural choice; it positions the two labs on opposite sides of a debate about what standards should govern AI contributions to mathematics. Whether peer review is sufficient to validate AI-generated proofs, or whether additional scrutiny is needed, remains genuinely contested among mathematicians and AI researchers.

None of the three camps, Pro-AI, Anti-AI, and Middle Ground, had published reactions at the time of writing. Pro-AI voices would typically treat results like these as strong evidence that AI systems are crossing into genuine scientific contribution rather than mere pattern-matching, and would likely use the peer-reviewed publication as a rebuttal to critics who question the rigor of AI-generated mathematics. Anti-AI voices would be expected to press on questions about interpretability, whether the proofs are human-verifiable in practice, and what it means for the discipline if AI systems begin closing long-standing open problems at scale. Middle Ground commentators would probably distinguish between the result itself and the process, welcoming the peer-review norm DeepMind established while calling for clearer community standards before such work reshapes research priorities.

The argument most worth watching is whether the mathematics community formally codifies peer review as a minimum bar for AI proof publication, and whether OpenAI responds to its boycott by changing its release practices. Any statement from major mathematics institutions or journals on those standards would move this debate significantly.