DeepMind Researcher Quits Over AI Safety Concerns

Robert O'Callahan spent years building chip design tools at Google DeepMind's New Zealand office, work that made AI models cheaper and faster to run. He just quit, and he's not staying quiet about why. In a public resignation post, O'Callahan says the industry's "current rate of change is far too high" and that he can no longer justify the contribution he made to speeding it up.

He's not the first DeepMind employee to walk out the door citing safety concerns, and he says he won't be the last. O'Callahan claims many of his former colleagues privately share his worries about the pace of AI development but almost never say so in public. That silence, he suggests, is part of the problem.

His departure lands at a moment when the debate over AI risk has become tangled up with politics. Effective altruism, the movement associated with authors like Eliezer Yudkowsky, has drawn fire from critics who accuse it of prioritizing charity over redistribution, and Axios reports a White House memo recently painted the movement as cult-like while singling out Anthropic CEO Dario Amodei as the face of AI "doomerism."