Cases in which the originators of technical works have come to suspect that AI platforms have appropriated those works as theirs continue to mount.
In September 2026, in a case that attracted global news media, a professor from New York University had been using OpenAI to develop a solution for a famous mathematics problem, which OpenAI later claimed to have solved itself.
In another case, described in November 2025, an AI app developer developed a concept which would make certain AI processes more effective – and two days later, it appeared as a feature in ChatGPT.
Concerns have not been limited to commercial interests. Also in September, US intelligence and security agencies announced details of “systematic extraction of proprietary functionalities and capabilities threatening U.S. technological leadership,” by China-based AI models.
Millennium Problem
In the case where NYU professor Tristan Buckmaster had been using OpenAI to develop a solution for the Navier-Stokes problem, one of seven “Millennium Problems” identified in the year 2000 by the Clay Mathmatics Institute.
When Prof. Buckmaster read OpenAI’s announcement that its platform had solved this problem, he noted that OpenAI was “…openly admitting they used training data from a period after we found our result.”
On further investigation, Prof. Buckmaster posted: “…I asked whether the model had been trained on, or had access to, our sessions in Codex, into which we had been putting all our drafts for the whole of this project. I was told the model did not look up user data. I asked again, about training, and I did not get an answer…”
Stolen technique?
The Navier-Stokes case wasn’t the first time that OpenAI was accused of appropriating a technical solution. In late 2025, Klaudi, an AI app developer, found that a concept which would make AI processes more effective. Two days later, the same technique appeared as an option in OpenAI’s Chat GPT.
While the technique was less efficient, that wasn’t the point. “The HugstonOne team celebrated the internal success of our memory toggle feature. We had a working prototype, a clear technical advantage, and a plan to roll it out. Then 2 days later a new, sleek interface on the ChatGPT platform,” said Klaudi.
“The UI is different, the branding is theirs, but the core concept the very idea of giving the user an on/off switch for LLM level memory is ours,” Klaudi said.
Why it matters
It’s massively difficult to determine whether or not acts of intellectual property theft have occurred. OpenAI says that the Navier Stokes work was theirs.
In an analysis published by The New York Times, “Although OpenAI now says it has determined Dr. Buckmaster’s research could not have been used to train its new system, the larger question remains. Today’s A.I. systems span trillions of “parameters” — the patterns they identify as companies like OpenAI train them on vast troves of data — and there is no way of combing though all these patterns to determine how a system ultimately found an answer.”
In the November 2025 Klaudi case, “Whether through data scraping, a compromised account, or a more direct form of corporate intelligence, the result is the same. The innovation, born of countless hours of hard work and investment, is being co opted without credit or compensation. ‘It’s not fun. It’s demoralizing and counterproductive,’ says a teammate on the project, who asked to remain anonymous.”
Now, the unlicensed use of intellectual property by nation-state actors adds to these concerns. Certainly cases like these (and the many cases – and lawsuits – in the media and entertainment industry space) will continue to bear close scrutiny.
Further reading
On the Navier-Stokes Millennium Prize Problem. Press release. September 8, 2026. OpenAI.
The Navier-Stokes equation. Informational landing page. Accessed Sept 9, 2026. Clay Mathmatics Institute.
‘…they are openly admitting…’ Social post by Tristan Buckmaster. Sept 8, 2026. Posted on Mastodon.social
‘…I asked whether the model had been trained on, or had access to, our sessions in Codex…’ Social post by Tristan Buckmaster. Sept 7, 2026. Posted on Mastodon.social
OpenAI says it has cracked one of math’s ‘Millennium Problems.’ Article. September 8, 2026. The New York Times
Is ChatGPT and OpenAI stealing ideas? Does it have the right to do so? How do we know it doesn’t? Article. November 15, 2025. by Klaudi. Published by Medium
