Video summary

OpenAI's biggest math breakthrough is getting ugly...

Main summary

Key takeaways

News and Commentary

Overview

OpenAI has reportedly claimed it “solved” the Navier–Stokes equations, one of the Millennium Prize Problems, triggering both excitement and controversy in the math community.

Navier–Stokes equations are widely used to model fluid and gas behavior, including applications such as:

  • hurricane forecasting
  • aircraft design
  • blood flow modeling

The Millennium question asks whether these equations can ever “break”—meaning they develop singularities or otherwise fail under certain conditions.

The Millennium Prize Problem (Background)

  • The Clay Mathematics Institute (CMI) offered a $1 million prize for either:
    1. Proving the equations never break, or
    2. Finding a counterexample showing they do break.

Prior Near-Success

Before OpenAI’s alleged breakthrough:

  • Tristan Buckmaster (NYU) and Levent Alpagay (Anthropic) reportedly worked for about a year on related progress.
  • Their prior work achieved a near milestone: showing a simpler related system—the “Euler equations”—could fail.
  • The story dates this earlier result to Aug 15.

OpenAI’s Alleged Rapid Breakthrough

In late August, the reporting claims OpenAI:

  • trained a new model using all unsolved Millennium problems at once
  • used approximately ~10,000 agents
  • spent around $20 million in compute
  • then claimed this produced a Navier–Stokes breakthrough

The video further alleges OpenAI’s approach closely matches the method Buckmaster and Alpagay developed.

Dispute Over Credit, Collaboration, and Influence

The reporting describes tensions around attribution and whether earlier work influenced OpenAI’s results.

Buckmaster’s allegations (timeline highlights)

  • Sept 3: Buckmaster emailed OpenAI to ask what was going on and later joined a call with Sebastian Bubeck.
  • Buckmaster alleges OpenAI’s team contributed significantly, including:
    • prompt writing by Codex
  • He also questions whether the model was trained on his earlier sessions.

Alleged publishing/treatment options

The video claims Buckmaster was presented with two options:

  1. Publish first on Euler results, while OpenAI publishes Navier–Stokes and credits Buckmaster as the “closest”
  2. Buckmaster could write the Navier–Stokes paper himself, but Alpagay would be excluded because he works at Anthropic

Buckmaster reportedly:

  • refused both options
  • threatened to go public
  • claims the interaction became tense

OpenAI’s competing explanation

OpenAI disputes Buckmaster’s account, stating:

  • they did not ask to remove Alpagay
  • they characterize their own actions as acting with integrity
  • the video also claims OpenAI portrayed Buckmaster as the one making threats

Public Statements and Publications

The reporting outlines a sequence of public materials:

  • Tuesday morning: Buckmaster and Alpagay posted their papers plus a four-page statement describing what they say happened.

  • Tuesday afternoon: OpenAI posted its own solution and a blog post, emphasizing:

    • no user data was accessed
    • the proofs are significantly different
    • That night: another commentator posted an additional statement portraying the situation as escalating in tone.

Broader Community Concern

The story cites concern from Terence Tao, warning that if mathematicians believe rumors about their work will trigger AI “agent swarms” that front-run them, researchers may stop sharing ideas—potentially damaging the culture of open science.

If mathematicians stop sharing ideas because of fear of being front-run, open science could be harmed.

Meta-Lesson Emphasized

The narrator frames the situation as a lesson about which companies to trust, tying it to the video sponsor.

Presenters or Contributors

  • Code Report (video host/narrator)
  • Tristan Buckmaster (NYU)
  • Levent Alpagay (Anthropic)
  • Sebastian Bubeck (OpenAI researcher)
  • Sam Olman (as named in the subtitles; appears intended to refer to Sam Altman)
  • Claude Navier (historical figure referenced)
  • George Stokes (historical figure referenced)
  • Terence Tao (quoted/cited take)
  • Madame Zerone (named in a late statement)

Original video