OpenAI agent does it again, solving nearly 400 math problems no human has ever solved

Summary
This news report by Nathaniel Dove for Global News covers OpenAI’s release claiming its unreleased AI agents solved 372 families of previously unsolved mathematical problems. It features commentary from Western University computer scientist and mathematician Mark Daley, examining both the mathematical community’s pushback regarding human insight and the potential broader benefits across science and medicine.

What is shown

• [00:00] Sam Altman speaking at an OpenAI event in San Francisco on September 29, 2026.
• [00:14] The document OpenAI Research Catalog: Mathematics (dated October 6, 2026), detailing 372 result families in 719 manuscripts across subjects like number theory, geometry, and analysis, highlighting problems such as the quasi-Riemann hypothesis and Kobayashi’s canonical-ampleness conjecture.
• [00:21] Video interview clips of Mark Daley discussing the shock value and style of the AI-generated proofs.
• [00:29] OpenAI simulation graphic depicting vortex dynamics related to an unraveled version of the Navier–Stokes equation.
• [00:40] Graphic of the September 11, 2026 open letter A Severe Misalignment of AI in Mathematics, signed by numerous Fields Medalists.
• [01:33] Statement by the Advisory Group on Mathematics and Artificial Intelligence (On OpenAI’s Release of Mathematical Results, October 8, 2026).
• [01:49] Visual of the 3D Navier–Stokes equations from NASA Glenn Research Center, followed by B-roll of financial markets, industrial manufacturing, and high school classrooms.

Claims & numbers

• The narrator reports that OpenAI claims its unreleased AI agents solved nearly 400 unsolved mathematical problems (specifically 372 result families across 719 manuscripts).
• The narrator states that OpenAI previously solved a version of the Navier–Stokes equation that had stood unsolved for decades.
• The narrator notes that about two dozen highly decorated mathematicians responded on September 11, 2026, with a letter arguing that problem solving is merely a proxy for conceptual understanding.
• Mark Daley states that many of the AI’s proofs are structured in a way that feels awkward to human mathematicians.
• The report highlights a statement from the Advisory Group on Mathematics and Artificial Intelligence dated October 8, 2026, framing the release as the beginning of incorporating AI work into human mathematical understanding rather than its completion.

Notable quotes

• [00:01] Sam Altman: “I believe that the future, if we get this right, can be much more like a new Renaissance...”
• [00:21] Mark Daley: “Releasing all of them at once definitely has shock value for the community.”
• [01:01] Mark Daley: “I’m going to grieve that as a human who wants to understand, but if that breakthrough leads to a new type of drug that cures 80% of cancers, it’s okay that I don’t understand it.”

Assessment
This is a standard broadcast news segment combining file footage, document excerpts, and an expert interview. The report does not verify or demonstrate any mathematical proofs firsthand, instead summarizing public announcements and the academic reaction surrounding OpenAI’s catalog release.

Described by gemini-3.8-flash on 2026-10-10 from the video’s audio and frames.