AI giant, Anthropic is calling on frontier AI developers to establish a coordinated, verifiable system for slowing down or temporarily pausing AI development if advanced systems begin improving themselves faster than society can manage the risks.
The AI company made the disclosure in a detailed post published on its website, warning that artificial intelligence capable of building its own successors could cause humans to lose meaningful control over the technology, and that the industry needs agreed-upon rules for when and how to hit the brakes before that threshold is reached.
The urgency behind the call is emphasized by a figure Anthropic disclosed alongside the announcement: as of May, more than 80% of the code merged into its own codebase was written by Claude, its flagship AI model, a sign that AI-assisted development has already moved well beyond a theoretical concern.
The company said a meaningful pause in frontier AI development would require participation from multiple well-resourced laboratories operating at the technological frontier, not a single company acting alone.
“It would be good for the world to have the option to slow or temporarily pause frontier AI development to enable societal structures and alignment research to keep up with the advance of the technology,” the company said.
“If systems are capable of fully building their own successors, the ways we secure them, monitor them, and shape their behavior all grow much more important,” it added.
Anthropic cautioned that a unilateral slowdown by one company would have limited impact, primarily shifting technological leadership to less cautious actors rather than producing the broader global deliberation the situation requires.
It said any credible pause mechanism would need clear rules on what conditions would trigger or lift a pause, which organisations would be involved, and who would have oversight authority.
“Full recursive self-improvement also might increase the risks of humans losing control over AI systems,” the company warned, describing the scenario as one of the most consequential developments in the history of technology.
Anthropic’s research arm, the Anthropic Institute, plans to study and help design the systems that would be necessary to support a coordinated slowdown if one were needed.
In the coming months, the company plans to convene a series of discussions bringing together policymakers, researchers, civil society organisations, and other AI firms to examine the key questions the proposal raises.
Those discussions will focus on how to manage risks associated with recursive self-improvement and how to strengthen coordination mechanisms across the industry.
The company acknowledged that poorly coordinated or unilateral slowdowns could backfire if less cautious actors continue advancing in the absence of a broader agreement, potentially reducing overall safety rather than improving it.
Earlier, in April 2026, Anthropic CEO Dario Amodei warned that artificial intelligence (AI) could replace a significant number of entry-level white-collar jobs within the next one to five years, raising concerns about potential job losses in industries such as finance, consulting, and technology.
Speaking in an interview with Fox News, Amodei said AI systems are advancing at a pace that many individuals, businesses, and policymakers may be underestimating.
According to him, tasks traditionally handled by junior employees, including document summarisation, idea generation, and financial report preparation, are increasingly being performed by modern AI systems.
