Robert O’Callahan, a technical expert at Google DeepMind in New Zealand, has resigned because he believes AI development is moving too quickly. He worked on chip-design tools that can make AI systems cheaper and faster, but said he could no longer justify contributing to that acceleration.
O’Callahan called the risk from superintelligent AI real but uncertain and argued that deliberately pursuing it in the near term is inherently irresponsible. His concerns extend beyond catastrophic scenarios to cognitive dependence, mental-health effects, concentrated power, cybersecurity, economic disruption and weak accountability. He said colleagues share some of these worries but are reluctant to discuss them publicly.
The resignation is one individual’s judgment, not evidence that a specific DeepMind system is unsafe. O’Callahan also stopped short of endorsing the claim that superintelligence would inevitably cause catastrophe. His departure nevertheless adds to a recurring governance problem for frontier labs: employees who help improve core capabilities may have few options beyond internal objections or resignation when they disagree with the organization’s pace. He plans to seek work he considers unambiguously beneficial to people.