DeepMind Technical Expert Robert O'Callahan Resigns, Citing Near‑Term Superintelligent AI as Irresponsible
The New Zealand‑based researcher leaves Google DeepMind, warning that the rapid pace toward artificial superintelligence poses real, uncertain risks that outweigh its benefits.
What happened
- On September 25, 2026, Robert O'Callahan, a technical expert at Google DeepMind’s New Zealand office, announced his departure from the company.
- He said the current velocity of AI development and the prospect of achieving superintelligent AI soon are “inherently irresponsible,” prompting him to seek “unambiguously pro‑human” work.
Details of the resignation
- O'Callahan’s résumé includes work on chip‑design tools that helped make AI models cheaper and faster—a contribution he now feels he can no longer justify.
- He characterizes the danger from artificial superintelligence (ASI) as “real but uncertain” and argues that aiming for ASI in the near future is morally untenable.
- Specific risks he highlighted:
- Cognitive surrender – humans relinquishing critical thinking to machines.
- AI‑induced psychosis and loneliness – mental‑health impacts tied to advanced AI interaction.
- Concentrated power – a few actors gaining outsized control over AI capabilities.
- Economic disruption – upheaval in labor markets and broader economic structures.
- Cybersecurity threats – new attack vectors enabled by powerful AI.
- Lack of accountability – difficulty assigning responsibility for AI‑driven outcomes.
- He believes the potential harms of AI outweigh its benefits and intends to focus on work that is clearly aligned with human well‑being.
- In his statement, O'Callahan linked to the essay “If Anyone Builds It” by Eliezer Yudkowsky and Nate Soares, though he stopped short of endorsing its claim that superintelligent AI inevitably leads to catastrophe.
- The essay’s authors are prominent within the effective‑altruism movement, a group that has faced criticism for emphasizing charitable giving over broader wealth redistribution.
- Axios reported that a recent White House memo described effective altruism as a “cult‑like fringe” and singled out Anthropic CEO Dario Amodei as a leading figure of AI “doomerism.”
- O'Callahan notes that many of his DeepMind colleagues share his concerns, but few have spoken publicly about them.
Industry context
- O'Callahan’s exit follows at least one earlier DeepMind researcher who left over AI‑risk worries, suggesting a growing internal unease about the speed of AI progress.
- The resignation adds to a wider conversation in the tech sector, where executives from firms such as Anthropic and OpenAI have been labeled “doomers” by political actors, reflecting heightened scrutiny of AI safety narratives.
