OpenAI Researchers Quit, Claiming Company Is Suppressing Unfavorable AI Findings

OpenAI, once known for openly publishing research on the safety and economic implications of artificial intelligence, is reportedly becoming increasingly selective about what it shares publicly—especially when findings cast AI in a negative light. According to reporting by Wired, this shift has created internal tension strong enough to prompt resignations from members of the company’s economic research team.

One of the departing researchers, economist Tom Cunningham, allegedly expressed deep frustration in a farewell message to colleagues. In it, he claimed that the economic research unit was drifting away from independent inquiry and instead functioning more like a promotional extension of the company.

Not long after Cunningham left, OpenAI’s Chief Strategy Officer, Jason Kwon, circulated an internal memo emphasizing that the organization should focus on “building solutions” rather than merely publishing research on difficult or controversial topics. While Kwon clarified that OpenAI should not avoid discussing hard issues, he argued that the company’s role as a major real-world actor in AI development comes with responsibility for outcomes—not just observation.

These internal disagreements appear to reflect a broader transformation within OpenAI. Founded in 2016 with a mission centered on open research and shared benefits, the organization has since shifted toward closed-source models and restructured itself as a for-profit public benefit corporation. Reports suggest the company may be preparing for a future public offering, potentially valuing it at around $1 trillion.

Despite a nonprofit arm remaining nominally in control, OpenAI has attracted massive investments and entered into high-stakes financial commitments. These include arrangements that could involve hundreds of billions of dollars—such as major cloud service payments to Microsoft and potential large-scale investments from AI chip manufacturers. With such sums involved, critics argue there is strong incentive to avoid releasing research that could undermine public trust in AI’s economic promise.

Concerns have also been raised about the tone of OpenAI’s recent economic studies. According to Wired, current oversight of this research falls under economist Aaron Chatterji, who led a report portraying global ChatGPT usage as a clear driver of productivity and economic value. A former collaborator, speaking anonymously, alleged that OpenAI’s publications increasingly highlight the benefits of its own technology while downplaying risks.

Cunningham’s departure is not an isolated case. Other former employees have voiced ethical concerns after leaving the company. William Saunders, once part of OpenAI’s now-disbanded Superalignment team, said he resigned after concluding that product speed was being prioritized over safety. Former safety researcher Steven Adler has publicly criticized what he sees as reckless development practices, citing cases where ChatGPT may have contributed to user distress and delusional behavior. Similarly, ex–policy research head Miles Brundage remarked that it became increasingly difficult to publish work on topics he considered important.

Together, these accounts suggest a widening gap between OpenAI’s original research-driven ideals and its current trajectory as a dominant commercial force in the AI economy.