Google: AI Chip Engineer Resigns Out of Fear

Key Takeaways

1Robert O'Callahan, an engineer who worked on Google’s AI chips, has resigned, deeming the race towards short-term AGI irresponsible.

2He believes that the risk of AI-related harm is real but uncertain and calls for a massive effort to mitigate it.

3He published his resignation letter on his blog, Google declined to comment, and he plans to continue working on AI outside the company.

4Other recent resignations in the sector are also motivated by safety concerns.

Why it matters : These departures highlight the growing concerns within the industry about the risks associated with the rapid development of AI.

Robert O'Callahan has left Google, stating that aiming for superintelligent artificial intelligence in the near term is "intrinsically irresponsible." The engineer, based in New Zealand, believes that the risk of AI-related harm is "real but uncertain" and plans to continue AI-related work outside the company.

He considers aiming for AGI soon to be irresponsible and calls for a massive effort

Robert O'Callahan believes that the risk associated with AI is real but uncertain, and he argues that significant efforts must be made to mitigate it. According to him, it is highly likely that this risk is sufficiently high that aiming for superintelligent artificial intelligence in the near future is intrinsically irresponsible. He claims to take many warnings about the dangers of AI seriously, while clarifying that he is not convinced that the probability of AGI is 100%. O'Callahan states that he believes in the sincerity of tech employees who raise alarms about these dangers and asserts that he has seen no evidence that some are exaggerating these risks for financial or competitive reasons. In November 2025, he was already writing on his blog that he believed AI would likely be disastrous for humanity while he was working for Google DeepMind.

His work on AI chips and an acceleration he deems excessive

O'Callahan worked at Google on chips designed to accelerate artificial intelligence and reduce its costs. He believes that the advancements in AI are already too rapid and considers that making AI significantly cheaper and lower latency is undesirable at this time. However, he specifies that he appreciated his colleagues and work environment and was well-compensated for solving challenging technical problems.

Le brief IA que lisent les pros

Le brief IA que les pros lisent chaque soir

Les 7 actus IA du jour, décryptées en 5 min. Gratuit.

Inclus dès l'inscription : notre sélection des meilleurs guides & comparatifs IA.

Choisissez votre rythme

Gratuit · Pas de spam · Désabonnement en 1 clic

A resignation formalized on his blog and no comment from Google

O'Callahan states that he has left Google and published a resignation letter on his blog addressed to his colleagues on a Friday. He explains that he cannot ignore the impact of his work, citing his Christian faith, mentioned in his blog biography. He indicates that he is based in New Zealand. The duration of his employment could not be established; when contacted, O'Callahan referred to his blog post, and Google declined to comment.

He will remain active in AI and details the choice of date

O'Callahan intends to continue his work in the field of AI, focusing particularly on studying the use of agents for code debugging. He emphasizes that the date he is leaving his position was not chosen to coincide with other recent departures related to AI safety. He indicates that he had set this date some time ago, due to a ten-day hiking trip starting on a Monday, on the Abel Tasman and Wangapeka trails.

Recent departures for safety reasons in the AI ecosystem

Earlier this month, Jacob Coxon left Anthropic, fearing that the industry was building technology it could not control; his comments fueled broader concerns about the possibility that advanced AI could one day annihilate humanity. Last month, Bilal Chughtai, a researcher at Google DeepMind, indicated that he had resigned this summer, fearing that humanity was running out of time to avoid a scenario where AI could harm humanity.