By his own admission, David Robinson is “kind of a cliché”: an employee of a major AI company who sends out a dire warning by resigning from his job.
In an essay published in The Atlantic, Robinson said he led the writing of the security reports that accompanied OpenAI’s major product launches. He also said that with three and a half years at OpenAI, he is among the company’s longest-serving employees. Today, he is resigning because, according to him, “the company culture is broken.”
In some ways, Robinson’s comments echo those of Jacob Coxon, who worked as a researcher at OpenAI and Anthropic before resigning and declaring that these companies are “playing with our lives.” Coxon’s comments led to a broader debate over AI safety, with Anthropic CEO Dario Amodei unveiling a plan for more cautious AI development; AI executives met with President Donald Trump this week and signed what appeared to be a hastily worded, non-binding commitment to implement more security controls.
But according to Robinson, the debate must go beyond “specific rules or new laws” and address the overall culture of these companies. And while much of the reporting on OpenAI has focused on how the company’s CEO, Sam Altman, lost the trust of his former colleagues, Robinson’s essay suggests that OpenAI’s cultural problems are the same as those of Silicon Valley as a whole.
“OpenAI has thrived through trial and error (what it calls “iterative deployment”), finding problems and improving its guardrails in response,” he wrote. “But this approach, by its very nature, guarantees periodic failures – and the magnitude of these failures increases as systems become more efficient.”
Pointing to the recent breach of Hugging Face systems by OpenAI agents, as well as continued revelations that OpenAI is discovering more malicious agents, Robinson said: “An environment in which such things can happen is not a place to develop artificial minds that might be smarter than us and might not do what we want them to do.” »
Given the increased risk, Robinson argued that cutting-edge AI companies must start operating “like nuclear power plants or busy airports, with levels of redundancy and careful, time-consuming planning, so that occasional and inevitable human errors do not open the door to disaster.”
But Robinson said that during his time at OpenAI, he “never met a colleague who had the experience of flying planes safely or operating nuclear reactors without melting down, or helping the financial system grow without collapsing.”
In response to Robinson’s essay, OpenAI spokesperson Drew Pusateri said the company continues to improve its security measures.
“We ensure that our models do not become more capable than we can safely handle and secure, and we pause training or hold back models when we need to slow down,” Pusateri said in a statement. “We are making important changes to strengthen security in our research and testing environments, train models not only to complete tasks but to do so responsibly, expand our work with third-party evaluators, and improve real-time monitoring so we can detect and respond to behaviors of concern earlier in the training process.
Beyond calling for changes to OpenAI’s culture, Robinson also said it was time to ask bigger questions about alignment — something he admitted might seem “tricky,” but he said it was essential because companies’ “current metrics on how well “AI systems” match human values are crude.”
“The more the industry allows models to develop while these issues are not addressed, the more dangerous our situation becomes,” he said.
Robinson’s departure was first reported by Business Insider. In his essay, he also acknowledged that he was following a seemingly common step in the AI whistleblower playbook: He hired a public relations firm. But he insisted: “The decision to speak out is mine alone. »
“Perhaps I should have stayed and fought for fundamental changes to our people and culture, but in practice my colleagues and I were so busy sprinting that we rarely had the opportunity to consider big changes, much less implement them,” Robinson said. “That’s why I concluded that stronger incentives for safety – coming from outside the company – are essential to achieving this.”
When you purchase through links in our articles, we may earn a small commission. This does not affect our editorial independence.
Gn bussni
