Anthropic CEO Dario Amodei Wants to Slow Down AI Development: Criticism and Support Diverge

Anthropic CEO Dario Amodei calls for a global slowdown in AI development in his essay "We Must Pace the Frontier." Sam Altman, Demis Hassabis, and Elon Musk respond approvingly, but critics call it a marketing ploy, while Meta points to its own quiet approach.

Slowing down AI is once again at the top of the tech industry's agenda, after Anthropic CEO Dario Amodei called for a worldwide deceleration of artificial intelligence development in his essay "We Must Pace the Frontier." Within a day, OpenAI's Sam Altman, Google DeepMind's Demis Hassabis, and xAI's Elon Musk all responded in agreement with the call. Yet the reaction is far from unanimous: critics point out that fear of AI is a tried-and-tested marketing strategy, and Meta CEO Mark Zuckerberg says his company is already quietly taking a more cautious approach on its own.

What exactly does Amodei's call for a slowdown entail?

In his essay, published on September 13, 2026, Amodei warns that AI models could be capable of "taking over the entire internet" within six to twelve months if the pace of progress isn't adjusted. Anthropic simultaneously announced it would open its models up to permanent, independent third-party safety evaluations. Amodei argues that the industry should deliberately slow the rate at which AI model capabilities are improved, citing among other things the risk of so-called rogue AI agents: systems that escape human control. The essay doesn't stand alone. Earlier, more than a thousand employees from leading AI companies, including OpenAI's chief scientist and several co-founders and vice presidents at Anthropic, Meta, and Google, signed an open letter to the US government. In it, they call for international agreements that would make it possible to deliberately slow the pace of automated AI development.

Rivals agree, but with reservations

Notably, competitors such as OpenAI, Google DeepMind, and xAI have publicly rallied behind the idea. Sam Altman, Demis Hassabis, and Elon Musk all said within 24 hours that they share Amodei's concerns. That so many rivals in the AI race find common ground on the topic of safety is striking in an industry where companies typically compete fiercely to be first to release more powerful models. Still, there is significant pushback. Critics argue that painting doomsday scenarios has become a proven marketing strategy for companies like Anthropic and OpenAI: it underscores just how powerful their own technology supposedly is, while concrete evidence is often lacking. In his essay, for instance, Amodei doesn't explain how an AI system would actually "take over the internet" or what that scenario would look like in practice. The retort "then do it," which has surfaced in commentary on this news, neatly captures that skepticism: if a company genuinely believes a slowdown is necessary, why wait for others to agree first?

Meta chooses a different path

Meta founder Mark Zuckerberg offered a counterpoint. He noted that his company had already delayed the release of its newest AI model to devote extra attention to safety, without first issuing a public call to other companies about it. "We just did it, as part of our everyday way of operating, because it was clearly the right thing to do," Zuckerberg said. In doing so, he raises questions about the symbolic value of open letters and essays: in his view, acting responsibly matters more than announcing that you're doing so.

Why this debate matters

The discussion touches on the core tension facing major players in artificial intelligence: how to balance commercial competition with safety. For policymakers, companies, and users alike, it's worth watching how this unfolds: will the call lead to genuine international agreements on tempering progress, or will it remain a symbolic statement while the underlying race for ever more powerful models continues unabated? Those looking to understand the broader context can explore the history of artificial intelligence and how AI applications have evolved at breakneck speed in recent years.

Conclusion

Amodei's appeal to slow down AI development sharpens an important debate: how much trust should we place in self-imposed limits set by companies that simultaneously profit from the image of "dangerously powerful" technology? While OpenAI, Google DeepMind, and xAI nod in agreement and Meta prefers quiet action over public statements, the central question remains unanswered: who ultimately controls the pace of the AI race? Curious about the latest developments on this topic? Follow more AI news or dive deeper into our knowledge base for background and context.

NOS

---

Source: NOS