ARTIFICE DAILY AI + CYBER SECURITY BRIEFING

Artificial Intelligence

Robert O'Callahan leaves Google DeepMind over AI risks

Robert O'Callahan said he left Google DeepMind because he could no longer justify work that helped make AI cheaper and faster, and he argued that near-term superintelligent AI was irresponsible.

THE-DECODER

Robert O'Callahan said he had left Google DeepMind, where he had worked as a technical specialist in New Zealand. He said he had helped build chip-design tools that made AI cheaper and faster, and he could no longer defend that contribution.

He said the pace of AI change was too high and described superintelligent AI as a real but uncertain risk. He warned about people surrendering judgement to systems, psychosis, loneliness, concentrated power, economic disruption, cybersecurity risks and weak accountability.

He said many colleagues at DeepMind shared similar worries but rarely spoke in public, and he said he now wanted to do work he saw as pro-human. He also referred to If Anyone Builds It by Eliezer Yudkowsky and Nate Soares without adopting its claim of inevitable catastrophe.

Named in this story

People

Robert O'Callahan
said he had left Google DeepMind over AI risk concerns
Eliezer Yudkowsky
co-author of If Anyone Builds It
Nate Soares
co-author of If Anyone Builds It

Companies

Google DeepMind
his employer and the source of the chip-design work he described

Products and systems

If Anyone Builds It
the book he linked to

Places

New Zealand
where he was based in the role

How the source tells it

The register was opinionated and advocacy-heavy, with weary repetition, alarmed risk language and an ethical framing of AI work.

  • weariness the opening treated the departure as another routine exit
  • alarm it stacked several severe AI harms and risks in one passage
  • moral framing it cast the work as a question of responsibility and pro-human purpose