David Robinson, a longtime OpenAI employee who led the writing of safety reports accompanying major product launches, has resigned from the company, arguing that its culture is “broken” and warning that current AI development practices are becoming increasingly risky as systems grow more capable. Robinson, who spent three and a half years at the company, said he is among OpenAI’s longest-serving employees and decided to leave after concluding that deeper changes were unlikely to happen internally.
His concerns were detailed in an essay published in The Atlantic. Robinson described himself as “something of a cliché” — an employee at a leading AI company who issues a public warning while resigning.
Debate Extends Beyond AI Regulations
Robinson’s departure comes amid an ongoing debate about AI safety. His comments echo concerns raised by former OpenAI and Anthropic researcher Jacob Coxon, whose resignation and warnings about advanced AI prompted wider discussion across the industry.
According to Robinson, the conversation should extend beyond new laws and specific regulations. He argued that the culture within frontier AI companies deserves greater scrutiny, adding that the issues he observed at OpenAI reflect practices common throughout Silicon Valley rather than problems unique to a single company.
Much of the public discussion surrounding OpenAI has focused on CEO Sam Altman and his relationships with former colleagues. Robinson, however, placed greater emphasis on organizational culture and decision-making processes.
Concerns Over Iterative Deployment
Robinson criticized OpenAI’s reliance on what the company calls “iterative deployment,” a process that releases systems, identifies problems, and strengthens safeguards afterward.
“OpenAI has thrived by trial and error,” Robinson wrote. “But this approach, by its very nature, guarantees periodic failures — and the scale of those failures is growing as systems get more capable.”
He pointed to the recent breach involving Hugging Face systems by OpenAI agents, along with reports that OpenAI continues to discover additional rogue AI agents. Robinson argued that such incidents demonstrate weaknesses in environments where increasingly powerful AI systems are being developed.
“An environment where things like this can happen is no place to grow artificial minds that could be smarter than we are and that might not do what we want them to,” he wrote.
Calls For Safety Practices Similar To High-Risk Industries
Robinson said frontier AI companies should adopt operating practices similar to those used by nuclear power facilities and major airports. He argued that these sectors rely on multiple layers of redundancy and extensive planning to prevent inevitable human mistakes from leading to severe consequences.
During his time at OpenAI, Robinson said he never encountered colleagues with professional experience in areas such as aviation safety, nuclear reactor operations, or maintaining financial system stability.
He also called for greater attention to AI alignment. While acknowledging that discussions about human values can sound abstract, Robinson argued that current methods used to measure whether AI systems align with human values remain limited.
“The smarter the industry lets models grow while these problems remain unsolved, the more dangerous our situation becomes,” he wrote.
OpenAI Says Safety Measures Continue To Expand
Responding to Robinson’s essay, OpenAI spokesperson Drew Pusateri said the company continues to strengthen its safety controls and is prepared to slow development when necessary.
“We’re making sure our models don’t become more capable than we can safely manage and secure, and we pause training or hold back models when we need to slow down,” Pusateri said.
In a statement published by OpenAI, Pusateri said the company is strengthening security within research and testing environments, expanding work with third-party evaluators, improving monitoring systems, and training models to complete tasks responsibly. He also pointed to OpenAI’s published report detailing how the company seeks to detect concerning behavior earlier in the training process.
Robinson’s resignation was first reported by Business Insider. In his essay, he acknowledged hiring a public relations firm, a step that has become increasingly common among former AI employees who publicly raise safety concerns, but insisted that the decision to speak out was his alone.
Reflecting on his departure, Robinson said the pace of work at OpenAI left little opportunity to pursue major organizational reforms. He said that although he considered staying and advocating for fundamental changes in staffing and culture, he ultimately concluded that stronger external incentives would be needed to improve safety practices.
Featured image credits: Wikimedia Commons
For more stories like it, click the +Follow button at the top of this page to follow us.