Key facts
- David Robinson, a former leader on OpenAI's Safety Systems team, has resigned.
- Robinson published an essay in The Atlantic detailing his concerns about AI safety.
- He stated that AI companies, including OpenAI, are "not being nearly careful enough."
- Robinson believes OpenAI's "culture is broken" and prioritizes "iterative deployment" over sufficient safety planning.
- OpenAI responded that it is strengthening security and pausing model training when necessary.
David Robinson, who led safety transparency at OpenAI, has resigned from the company, expressing concerns that AI developers are "not being nearly careful enough." In an essay published in The Atlantic, Robinson stated that OpenAI is "failing to achieve the level of care that I believe is needed" as it prioritizes rapid model development over safety.
Robinson, a Rhodes Scholar with a degree in philosophy from Princeton, joined OpenAI in 2023. He previously co-founded Upturn, an organization focused on using algorithmic technologies to improve government data accessibility, and lectured tech executives on ethics. He described his decision to speak out as following a common AI whistleblower playbook, but insisted it was his own choice.
His departure echoes similar warnings from other safety researchers, including Jacob Coxon, a former researcher at OpenAI and Anthropic. Robinson argued that AI companies should operate with the same level of caution as nuclear power plants or busy airports, emphasizing the need for layers of redundancy and careful planning to prevent catastrophic failures. He noted that he "never encountered a colleague who had experience making airplanes fly safely or nuclear reactors run without melting down."
In response to Robinson's essay, OpenAI spokesperson Drew Pusateri stated that the company is continuously strengthening its safety and security practices. "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. He added that OpenAI is making significant changes to strengthen security, train models to act responsibly, expand third-party evaluations, and improve real-time monitoring.
Robinson also called for a deeper examination of AI alignment, stating that current measures of how well AI systems match human values are "coarse." He warned that the smarter the industry lets models grow while these problems remain unsolved, the more dangerous the situation becomes. Robinson concluded that stronger incentives for safety from outside the company are crucial for addressing these issues.
