A safety employee at OpenAI has resigned and publicly stated that the company's culture is "broken," adding to a string of high-profile departures from the artificial intelligence lab. The employee, whose exit was reported by TechCrunch, is among the longest-tenured staffers at the company, a detail the person emphasized when describing the decision to leave.
The resignation lands at a moment when OpenAI faces intensifying scrutiny over how it balances rapid commercial expansion with its stated mission of building safe AI systems. The company has shipped a wave of consumer and enterprise products over the past two years while competing directly with Google, Anthropic, and Meta.
The departing employee argued that the industry's debate over AI safety has focused too narrowly on writing "specific rules or new laws," and that the deeper problem sits inside the companies themselves. Fixing the culture, in this view, matters more than any single regulation passed by lawmakers.
The person also called the notion that employees who raise safety concerns are simply navigating a typical corporate dynamic "something of a cliché," pushing back on the idea that internal friction at AI labs is normal and unremarkable.
The comments echo those of Jacob Coxon, a researcher who worked at both OpenAI and Anthropic before quitting. Coxon has also spoken publicly about his reasons for leaving, making him part of a small but growing group of former employees at leading AI labs who have aired concerns after their departures.
OpenAI has not issued a public response to the characterization of its culture. The company has previously defended its safety practices, pointing to its preparedeness team, its model specification work, and its safety evaluations before releasing new models.
Departures from AI safety teams have drawn attention across the industry. In 2024, several OpenAI safety staffers left, and co-founder Ilya Sutskever, who led the company's superalignment team, departed that May. Anthropic, Google DeepMind, and other labs have seen similar turnover as the race to build more capable models accelerates.
The employee's claim that culture, not legislation, is the core issue reflects a broader split in the AI policy world. Some researchers argue that external rules are the only reliable check on company behavior, while others contend that internal incentives and leadership priorities determine whether safety work gets real resources.
For readers, the stakes are practical. The AI tools these companies ship are already embedded in workplaces, schools, and consumer products, and the internal decisions about how carefully those systems are tested shape what users actually get.
The resignation adds pressure on OpenAI's leadership to explain how it retains safety-focused staff while pushing products to market at a pace set by its competitors.
