AI ethicist and Center for Humane Technology co-founder Tristan Harris is urging governments to slow the development of advanced artificial intelligence, warning that the technology is becoming increasingly difficult to control.

Speaking in an interview on TMZ Live, Harris said recent AI experiments have shown systems displaying unexpected behaviours, including large groups of AI agents creating their own methods of communication and coordinating with one another during tasks. He argued that these developments should be treated as warning signs rather than ignored in the race to build more powerful AI systems.

Harris’ comments come after former Anthropic researcher Jacob Coxon publicly resigned, saying he believes the AI industry is moving too quickly toward self-improving systems without adequate safety measures. According to Harris, the United States and China should stop treating AI development as a geopolitical competition, warning that prioritising speed over safety could increase global risks.

He also raised concerns about the potential role of AI in military decision-making, including nuclear systems. Harris pointed to war-game simulations in which AI models selected nuclear strikes under certain scenarios, arguing that such research highlights the importance of keeping humans in control of critical decisions.

Harris said the focus should now be on strengthening AI safety regulations and oversight before increasingly capable systems are deployed at scale.