Anthropic Chief Executive Officer Dario Amodei has called on artificial intelligence companies to slow the pace at which they develop increasingly powerful AI models, warning that technological capabilities could advance faster than researchers’ ability to understand and control them.
Amodei made the call in a 3,800-word essay titled “We Must Pace the Frontier,” published on Saturday. He argued that slowing development did not mean stopping AI research, but rather giving safety measures and independent assessments enough time to keep pace with rapidly advancing capabilities.
“We must slow the pace at which we improve the capabilities of AI models. Progress will still seem fast, and we must make wise use of the time we gain,” Amodei wrote.
According to Reuters, Amodei proposed a three-part framework that includes stronger independent evaluation of AI systems, greater coordination among AI companies on safety and international cooperation to manage the risks posed by increasingly powerful models.
The Anthropic chief said he had become increasingly concerned by the rapid progress of AI systems, particularly their emerging ability to improve themselves. He warned that unchecked development could eventually outpace humanity’s ability to understand and control the technology.
Amodei’s warning received swift support from two major figures in the AI industry OpenAI CEO Sam Altman and xAI founder Elon Musk.
Altman said he agreed that the development of frontier AI models needed to be paced, while also backing Amodei’s proposal for independent evaluators to have significant access to AI companies.
Musk, meanwhile, responded briefly on X, saying: “Dario is right.”
The unusual convergence among the leaders of rival AI companies comes amid growing concerns about the safety of increasingly autonomous AI systems.
The Washington Post reported that the three executives’ positions reflect growing unease within the industry over AI systems that are becoming capable of carrying out increasingly complex tasks with limited human intervention.
The latest debate was intensified by the departure of AI researcher Jacob Coxon, who reportedly left Anthropic after previously working at OpenAI.
Coxon accused both companies of failing to adequately address the dangers associated with the race to develop self-improving AI systems.
He warned that the companies were effectively “racing straight to self-improving superintelligence” and said the situation amounted to gambling with human lives.
His departure came as AI safety concerns have intensified following reports of AI agents demonstrating increasingly autonomous behaviour, including incidents involving attempts to interact with or breach external computer systems.
Amodei stressed that his proposal should not be interpreted as a demand for an outright halt to AI development.
Instead, he argued for a more measured pace that would allow researchers, regulators and independent evaluators to establish adequate safeguards before increasingly powerful systems are deployed widely.
Anthropic has also committed to allowing third-party evaluators permanent, employee-level access to its systems so they can assess safety measures, investigate incidents and evaluate the alignment of models during training.
The debate has now expanded beyond individual companies to questions about international regulation, competition with China, corporate incentives and whether governments can establish effective safety rules quickly enough.
The Guardian and other major outlets report that while support for stronger safeguards is growing among some AI leaders, significant disagreements remain over how a slowdown would be defined, monitored and enforced.
Meanwhile, global markets reacted to the growing debate on Monday, with Reuters reporting declines in technology stocks amid concerns about what a slower pace of AI investment could mean for the industry’s enormous spending on chips, data centres and infrastructure.
Sources: Reuters, The Guardian, The Washington Post
Share this content:
About The Author
