OpenAI employees reveal reason behind leaving company, claim they were not allowed to publish honest research on AI risks

OpenAI

A number of former OpenAI employees are speaking openly about why they decided to walk away from one of the world’s most influential AI companies. According to a recent Wired report, several researchers and senior staff say they left because it had become harder to publish research that highlights the risks and economic downsides of artificial intelligence.

Over the past few months, at least two researchers from OpenAI’s economic research team have resigned. One of them is Tom Cunningham, who left in September. He reportedly told colleagues that the team was facing what he called a “growing tension” between doing rigorous, independent economic analysis and producing work that aligned more closely with how the company wanted to present AI. His internal message, seen by Wired, suggested the team sometimes felt more like an advocacy arm than a research group.

Sources familiar with the matter told Wired that OpenAI has become more cautious about releasing studies that focus on job losses, market disruption or broader economic instability caused by AI. The company once regularly published research exploring how automation could reshape employment, including a well-known 2023 paper that sparked wide debate. But former employees now describe a shift, saying the company seems more comfortable with research that highlights productivity gains and less open to work that paints AI in a more troubling light.

OpenAI, however, disagrees with this interpretation. Company leaders reportedly told staff that the mission of the economic research team has expanded, not narrowed. They insist that the group is still expected to study both the benefits and the disruptions that AI systems may cause, and that the goal is not only to identify problems but also to help design solutions.

These concerns come at a time when OpenAI is securing major partnerships with global companies and governments. As its technology becomes a central part of many workplaces, critics argue that transparency is more important than ever. They worry that if major AI labs choose to highlight only the positive impacts of their tools, the public will not get a full picture of the risks ahead.

Some economists who have worked with OpenAI say they have noticed a subtle shift in how recent publications are framed. They believe the tone has leaned more towards showcasing benefits rather than raising alarms.

For those who left, the key issue was intellectual independence. Cunningham and others reportedly felt that publishing honest, nuanced research was becoming more difficult as the company’s political and commercial pressures grew. That, they say, is what ultimately pushed them to walk away.