4d ago3 sources0 reactions

Google DeepMind researcher quits over superintelligent AI safety concerns

35 DoomStory toneSafety concern, framed as principled resignation

By Splitfeed EditorialUpdated Sep 26, 2026, 19:00 UTC

3 sources · Google News · The Decoder · Business Insider

Doom: O'Callahan says building superintelligent AI soon is "inherently irresponsible" and the pace too high
Doom: He worked on chip design tools that made AI cheaper and faster, work he now rejects
Doom: Many DeepMind colleagues share his concerns but rarely speak out publicly, he says
Neutral: His is described as another in a series of safety-motivated departures from Google DeepMind

The story in full

Robert O'Callahan, a researcher at Google DeepMind, resigned on or around September 25, 2025, stating that building superintelligent AI soon is "inherently irresponsible" and that the "current rate of change is far too high." He had worked on chip design tools that made AI development cheaper and faster, work he said he could no longer justify.

O'Callahan noted that many of his colleagues share similar concerns but rarely speak out publicly. His departure is described as another in a series of exits from Google DeepMind by researchers with safety objections, suggesting internal unease that has not been widely aired.

Analysis

387 words

By Splitfeed EditorialLast updated Sep 26, 2026, 19:00 UTC

On September 25, 2025, Robert O'Callahan resigned from Google DeepMind, citing safety objections to the pace and direction of advanced AI development. O'Callahan had worked specifically on chip design tools designed to make AI training cheaper and faster, and he stated he could no longer justify that contribution. His core objection was direct: building superintelligent AI in the near term is, in his words, "inherently irresponsible," and the "current rate of change is far too high." He also noted that a significant number of his colleagues hold similar views but feel unable to say so publicly.

The resignation carries weight beyond one person leaving a job because of what it implies about conditions inside one of the world's leading AI laboratories. O'Callahan is described as another in a series of researchers who have left Google DeepMind over safety objections, suggesting a pattern rather than an isolated case. The fact that he characterizes internal concern as widespread but largely silent raises questions about whether dissent is being expressed through internal channels, suppressed informally, or simply crowded out by the competitive pressure to ship. What remains genuinely contested is whether the concerns he describes reflect a realistic near-term risk or a speculative one that should not drive decisions about research pace.

None of the three camps had published reactions by the time this story was filed, so what follows reflects the positions each would typically take. The Pro-AI camp would likely argue that safety-motivated departures are a sign that the field takes risk seriously, and that slowing development unilaterally only cedes ground to less safety-conscious actors. The Anti-AI camp would treat O'Callahan's resignation as direct evidence that insiders believe the industry is moving recklessly, and would point to the silence of his colleagues as a sign that institutional incentives suppress honest risk assessment. The Middle Ground camp would probably call for structured internal dissent mechanisms and independent oversight, arguing that neither full acceleration nor a halt is the right response to concerns like those O'Callahan raised.

The argument would sharpen considerably if more named researchers from Google DeepMind spoke publicly, or if the company responded to the pattern of safety-motivated departures with a formal statement on its internal risk governance. Any such response, or the absence of one, would give the competing camps more concrete ground to argue over.