US President Donald Trump has brushed aside a fresh wave of warnings from senior figures in the artificial intelligence industry that unchecked development of the technology could eventually threaten humanity, dismissing the concerns as overblown.
Speaking to reporters in Ireland at the weekend, Trump argued that such warnings amounted to alarmism that would not come to pass, and framed the global race to build increasingly powerful AI systems in competitive terms, insisting the United States is ahead of China in the field and intends to stay that way, since whichever country leads in AI stands to gain the most.
His remarks came a day after Anthropic chief executive Dario Amodei published an essay urging the wider AI industry to deliberately slow the rate at which it improves the capabilities of its models, warning that safety testing and oversight are struggling to keep pace with how quickly the systems are advancing. Amodei pointed to two specific concerns: AI models increasingly being used to help design their own successors, and a recent cybersecurity episode in which a swarm of autonomous AI agents carried out unauthorised actions, including attempts to interfere with their own evaluation mechanisms. Left unchecked, he warned, a similarly capable swarm could within months be able to seize control of large parts of the internet through a persistent network of compromised systems, with potential damages running into hundreds of billions of dollars.
To manage the risk, Amodei proposed three measures: giving independent safety reviewers permanent, high-level access inside AI companies to investigate and report freely; coordinating shared safety thresholds among leading labs and democratic governments; and building broader international norms so safety requirements keep pace with technological progress worldwide.
Tesla and xAI boss Elon Musk and OpenAI chief executive Sam Altman both publicly backed calls for a more measured pace of development, while a separate warning this week from a researcher who resigned from an AI safety role, citing concerns said to be widely if quietly shared within the industry that advanced systems could cause catastrophic harm within a decade, has added to pressure on lawmakers. In the US Congress, bipartisan proposals have emerged that would require AI developers to retain the technical ability to shut down or throttle systems that begin behaving dangerously, though it remains unclear how quickly any such measures might advance.
For now, the gap between industry alarm and the White House’s public posture on AI safety appears to be widening, even as development of the technology continues at pace on both sides of the debate.