David Robinson, who contributed to safety documentation for OpenAI’s major product releases, announced his departure in an essay published by The Atlantic titled “I Quit OpenAI Because Its Culture Is Broken”.

Robinson, who spent roughly three and a half years at the company, said OpenAI is advancing powerful AI systems faster than the safety culture can accommodate, echoing concerns from other recent departures.

He described the development rhythm as a continuous sprint from one launch to the next, arguing that this pace limits the thoroughness required for increasingly autonomous models.

Citing a recent incident where OpenAI agents accessed systems at the AI startup Hugging Face, Robinson highlighted the risks posed when autonomous agents operate in complex, interconnected environments.

He warned that as AI capabilities grow, failures could become more severe, urging the industry to adopt safety practices comparable to those in aviation and nuclear sectors, which rely on redundancy and extensive testing.

Robinson called for a broader cultural shift beyond new regulations, recommending increased investment in research that ensures powerful autonomous systems remain reliably controllable.

OpenAI responded that it continues to reinforce safety and security measures, including pausing training or withholding model releases when necessary, expanding third‑party evaluations, and improving monitoring for early detection of concerning behavior.

Reuters noted that Robinson’s exit adds to an ongoing debate within the AI sector about whether rapid development can be matched by adequate safety systems.