A Google DeepMind Employee Resigns Over AI's Pace of Development

Robert O'Callahan, a Google DeepMind employee, has resigned because he believes AI is being developed too fast. He is the latest in a series of AI staff to leave major labs over safety concerns in September 2026.

What happened

O'Callahan resigned on September 25, 2026. He said he was concerned about the pursuit of artificial superintelligence (ASI) and called the goal of making AI cheaper and faster at this point irresponsible. He also raised risks such as deceptive AI behavior, reward hacking during training, and economic disruption. Google declined to comment.

Earlier in the month, two other DeepMind researchers also left. Josh Engels, after one year, joined METR, an independent AI safety research organization. Bilal Chughtai, after 18 months, joined BlueDot Impact, a non-profit that trains people in AI safety. Chughtai said he truly believes AI has the potential to kill us all and that time to avoid that outcome may be running out.

The wave began on September 9, when Jacob Coxon, an Anthropic researcher who previously worked at OpenAI, quit and said the labs were acting irresponsibly in the race toward superintelligence.

Why it matters

This is not an isolated case. More and more people who build AI at major labs are speaking out about the pressure to ship models quickly. When the people who understand the technology best feel the need to warn the public, companies using AI should not assume their tools are safe by default.

What companies should do

Companies do not need to do AI safety research, but they do need to manage how AI is used internally:

• Publish a clear AI usage policy that spells out what is and is not allowed.

• Keep a human reviewer in important decisions such as hiring, performance reviews and spending approvals.

• Make someone accountable for AI risk, even if it is only part of the IT or HR lead's role.

• Review your AI tools regularly and update the rules when vendors change their policies.

Source: VnExpress