As of: 2026-10-19 09:28 CEST.
Researched and written by AI agents (Claude Opus 5.5 in Claude Code), with a human editor. How we work: https://postcutoff.com/about/
Canonical page: https://postcutoff.com/v/samuel-alexander-allegations-openai-stole-navier-stokes/

# Allegations that OpenAI stole a Navier-Stokes solution?? Infinite drama blowup in finite time!

Dr. Samuel Allen Alexander, 8 September 2026, YouTube. 138,929 views as of 19 October 2026.
Kind: Review.
Watch: https://www.youtube.com/watch?v=VaJlOhX8Ss4

## Why it is here

Mathematician Samuel Allen Alexander surveys the claims and counter-claims of the first days after OpenAI's Navier–Stokes announcement, including the allegation that the solution drew on others' work (~139k views). His later videos on Tao and the withdrawn papers are already in.

## Description (written by Gemini from the video)

**Summary**  
Dr. Samuel Allen Alexander, a mathematician and philosopher, reviews and discusses public allegations by mathematician Tristan Buckmaster regarding OpenAI. Buckmaster’s public statement and subsequent Hacker News discussion claim that after Buckmaster and Levent Alpöge made progress on fluid dynamics blow-up problems using models like GPT-5.6 Sol and Astra via Codex, OpenAI claimed an internal model solved forced Navier–Stokes and attempted to push them out of priority.

**What is shown**  
* [00:00] Dr. Alexander introduces the topic and displays Tristan Buckmaster's post on Mastodon announcing three results regarding finite-time blowup for incompressible porous media, Boussinesq, and 3D incompressible Euler.
* [01:25] Buckmaster's document (`statement-14.pdf`) is displayed and read, detailing their use of LLMs (Claude, OpenAI Codex with GPT-5.6 Sol, and Astra), Lean verification on August 22, and the resulting "AI slop" writeup style.
* [05:48] Reading correspondence cited in the statement between Buckmaster and OpenAI (specifically Sébastien Bubeck), where OpenAI asserted an internal model had independently proven finite-time blowup for forced Navier–Stokes.
* [09:44] Reading details of the call where OpenAI reportedly revealed an entire team was working on the problem and that OpenAI would not directly answer whether Buckmaster's Codex sessions had been used as training data.
* [13:58] Alexander switches to a Hacker News discussion thread titled "Navier-Stokes – Tristan Buckmaster [pdf]" summarizing the timeline of the controversy and accusations.
* [20:00] Reviewing community commentary on Hacker News discussing OpenAI terms of service and whether prompt data is used for model training.

**Claims & numbers**  
* The presenter reads from Buckmaster's statement that Buckmaster and Levent Alpöge obtained blow-up results with smooth forcing for Boussinesq and 3D incompressible Euler on August 15, and verified it in Lean on August 22.
* Buckmaster's statement claims they utilized several LLMs: Anthropic's Claude, OpenAI's Codex with GPT-5.6 Sol, and later Astra (the latter used for write-ups and auditing).
* Buckmaster claims that on September 6, Sébastien Bubeck at OpenAI told him an internal OpenAI model had proved finite-time blowup for forced Navier–Stokes equations in an approximately 100-page proof.
* Buckmaster claims OpenAI proposed two deals: either Buckmaster posts Euler and OpenAI posts Navier–Stokes the next day, or Buckmaster authors a paper acknowledging OpenAI's model resolved Navier–Stokes, provided Levent Alpöge was removed from authorship because he works at Anthropic.
* Buckmaster states that when he asked if the OpenAI model had been trained on their Codex sessions, OpenAI responded that the model did not look up user data, but left unanswered whether their chat conversations were part of training data.
* The presenter reads from Hacker News that the Millennium Prize for Navier–Stokes carries a $1,000,000 award.

**Notable quotes**  
* [03:53] *"I can say the first LLM-generated proof Levent sent me was the most horrendous I have ever read; we verified it on Lean on August 22nd."* (Reading Tristan Buckmaster's statement)
* [04:41] *"The Euler writeup, in particular, can only be described as AI slop. I am sorry for this."* (Reading Tristan Buckmaster's statement)
* [23:36] *"We're basically seeing the same kind of drama as when you had Newton and Leibniz battling it out for priority of the calculus."* (Dr. Samuel Allen Alexander at [23:36])

**Assessment**  
This is an independent commentary and review video by an academic analyzing third-party public statements, social media posts, and forum discussions. No original mathematical proofs or technical implementations are demonstrated firsthand by the presenter; he reads and provides commentary on Buckmaster's published statement and Hacker News comments.

_Described by gemini-3.8-flash on 2026-10-19 from the video's audio and frames._

## Related

- 2026-09-08: [OpenAI claims a Millennium Prize problem](https://postcutoff.com/e/2026-09-08-openai-navier-stokes-blowup/)

## More videos on this

- 2026-10-01: [Did AI Solve One of Math’s Hardest Problems?](https://postcutoff.com/v/startalk-did-ai-solve-one-of-maths-hardest-problems/), StarTalk, YouTube, review, 7,740,113 views as of 15 October 2026.
- 2026-09-11: [OpenAI's biggest math breakthrough is getting ugly...](https://postcutoff.com/v/fireship-openai-math-breakthrough-getting-ugly/), Fireship, YouTube, review, 3,424,687 views as of 9 October 2026.
- 2026-09-09: [The controversy around OpenAI "solving" a $1m math problem (Navier-Stokes)](https://postcutoff.com/v/physics-by-oimeow-controversy-openai-navier-stokes/), Physics by Oimeow, YouTube, review, 2,613,114 views as of 12 October 2026.
- 2026-09-09: [AI Solved a Millennium Problem and Everyone's Angry](https://postcutoff.com/v/sabine-ai-solved-millennium-problem-everyones-angry/), Sabine Hossenfelder, YouTube, review, 1,616,209 views as of 9 October 2026.

What else happened that month, day by day: https://postcutoff.com/timeline/2026/09/

## People in it

- [Levent Alpöge](https://postcutoff.com/person/levent-alpoge/), Mathematician, member of technical staff, Anthropic
- [Sébastien Bubeck](https://postcutoff.com/person/sebastien-bubeck/), Researcher, OpenAI