Ex-DeepMind scientist says AI is moving too fast to control

Alex Turner, a former research scientist on Google Alphabet Inc. ↗ 3.06% DeepMind's scalable alignment team, appeared on Bloomberg on Monday to argue that the pace of AI development has outstripped the ability of governments, companies, and rival superpowers to manage its risks. His remarks came amid an unprecedented wave of calls from top AI executives to slow the field's breakneck advance.

A Whistleblower's Growing Platform

Turner resigned from Google DeepMind in June after the company signed an agreement allowing the Pentagon to use its AI for classified work, a deal he said lacked binding restrictions against autonomous weapons or mass surveillance. Since then, he has become one of the most prominent voices warning about AI's trajectory. In a CNBC appearance last week, he said he was "concerned with extinction risk" from the technology.cnbc+2

In Monday's Bloomberg interview, Turner highlighted recursive self-improvement — the ability of AI systems to enhance their own capabilities — as a central danger, warning it could rapidly produce intelligence beyond human comprehension. He also pointed to compute tracking as a practical lever for regulation, arguing that monitoring the vast processing power required to train frontier models could give governments a concrete way to enforce limits. Turner acknowledged that Big Tech's recent calls to slow down may be partly sincere and partly a strategic move to consolidate advantage.bloomberg

Industry Chorus Grows Louder

Turner's warnings land during a pivotal week for the AI safety debate. On Saturday, Anthropic CEO Dario Amodei published a 3,800-word essay titled "We Must Pace the Frontier," urging labs to deliberately throttle capability gains so alignment research can keep up. OpenAI CEO Sam Altman and Elon Musk both publicly endorsed the message. Demis Hassabis, who leads Google DeepMind, also expressed agreement.cnbc+1

Amodei laid out a three-step framework: embedding independent evaluators inside labs, coordinating safety standards across democratic nations, and eventually extending those agreements globally — including to China. As of Monday, however, no rival lab had matched Anthropic's concrete commitment to grant third-party auditors employee-level access.shattered

The China Dilemma

The geopolitical dimension looms large. Amodei himself called China "the toughest dilemma" for any slowdown proposal, acknowledging the risk that unilateral restraint by Western labs could cede ground to Beijing. Turner, in a separate exchange on social media, pushed back against the Trump administration's AI czar David Sacks, who dismissed calls for pacing laws as an attempt to throttle competition. Turner wrote in a Guardian op-ed published Sunday that the solution is "straightforward: prevent companies from enabling AI to self-improve beyond a manageable level of intelligence".politomix+3