Allegations of Intellectual Property Misappropriation Surface Against OpenAI
OpenAI is facing serious allegations regarding the unauthorized use of proprietary research to train its large language models (LLMs). The controversy centers on claims that the company’s models, specifically Codex and its successors, may have ingested private mathematical research and intellectual property before it could be formally published.
Claims of Research Appropriation
According to reports circulating on social media platforms like Reddit, two mathematicians allege that OpenAI utilized their private work to develop solutions to complex mathematical problems. The researchers reportedly spent a year working on a difficult mathematical problem, feeding various drafts of their work into Codex. According to the researchers, OpenAI released solutions to the same problem via its “Sol” and “Astra” models just days before the mathematicians were set to publish their findings.
When questioned whether their models were trained on these private chats, OpenAI has reportedly declined to provide a definitive answer. While these claims currently lack independent verification from a formal legal or academic body, they have prompted significant debate regarding the privacy of cloud-based AI interactions.
The Navier-Stokes Controversy
The allegations extend to the realm of Millennium Prize Problems. Unverified reports suggest that OpenAI may have claimed credit for solving the Navier-Stokes equations, a feat that carries significant academic and financial prestige.
In one specific account shared via online discussions, a mathematician alleged that OpenAI engaged in “shady” negotiations regarding the publication of a solution. The mathematician claims OpenAI offered two options: either the researcher publishes his original discovery first, followed by OpenAI’s solution a day later, or the researcher publishes OpenAI’s solution under his own name while crediting the company to secure the prize.
Furthermore, a physicist has reportedly initiated legal action against OpenAI, claiming that the company used Codex to solve a problem that the physicist had previously solved using the same tool, only for OpenAI to claim the breakthrough shortly thereafter.
Data Privacy and Training Methodologies
The controversy highlights ongoing concerns regarding how AI labs utilize user data. Critics argue that even when users pay for API access, their inputs may be used to extract “hard facts or ideas” that are then integrated into training sets.
Industry observers note that while individual identities may be obscured through data aggregation, the underlying intellectual value—such as medical trends or mathematical breakthroughs—remains highly valuable for model training. This has led to questions regarding the transparency of “open-weight” models, as even models with accessible weights can be trained on proprietary, non-public datasets.
As of this report, OpenAI has not issued a formal rebuttal to the specific claims regarding the Navier-Stokes problem or the alleged misappropriation of the mathematicians’ drafts.