OpenAI safety leader David Robinson resigns, warning company culture is dangerous

David Robinson, a leader on OpenAI’s Safety Systems team, quit the company and followed with an essay criticizing its culture. Business Insider reported that Robinson left last week, adding to upheaval in OpenAI’s safety ranks. He worked on safety transparency, including helping develop and share system cards that provide information about OpenAI’s models.

Robinson argued that OpenAI’s “unimpeded optimism” has created a dangerous environment for powerful AI technology. He wrote that companies building the technology “aren’t being nearly careful enough” and said OpenAI is failing to meet the level of care he believes is needed as it moves from launch to launch. He framed the problem as cultural, saying the moment requires humility from people who have succeeded through “extreme confidence.”

OpenAI said it is “making sure our models don’t become more capable than we can safely manage and secure,” adding that it pauses training or holds back models when needed. The company also said it has expanded work with outside evaluators and is trying to improve real-time monitoring to detect and stop concerning AI behavior earlier in training. Its response positions safety controls as an active part of model development, even as Robinson questions whether the culture is moving fast enough.

Robinson’s resignation comes amid increased attention on OpenAI’s safety practices and staffing. The company recently said it had parted ways with three researchers over alleged violations of policies for sharing sensitive information, and safety head Johannes Heidecke also left earlier this year. The broader debate remains tense: some researchers and executives have called for slowing AI development, while others warn that slowing Silicon Valley could let international rivals and bad actors gain ground. OpenAI CEO Sam Altman said at the company’s developer conference that safety and alignment need to stay ahead of model capabilities.