Anthropic says AI is advancing so fast that leading labs may need to slow down.
In a blog post published Thursday, researchers at The Anthropic Institute, the company’s research arm, said that AI is already speeding up the development of new AI models so quickly it could eventually help build its own successors.
“We believe 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.
Anthropic’s warning comes as AI is reshaping the tech industry and the broader economy at breakneck speed.
In recent months, executives have touted AI’s growing ability to perform work once done by humans, from writing software code to handling recruiting and customer service tasks. Google has said AI now generates 75% of its code, while Mercor CEO Brendan Foody recently said that his startup spends more on AI tokens than employee salaries.
At the same time, a growing list of companies has linked layoffs and restructurings to AI-driven efficiency gains.
Anthropic pointed to its own internal data, showing how much work is already being delegated to AI.
It said that more than 80% of the code merged into its codebase is now written by Claude, and that the typical engineer was merging eight times as much code per day in the second quarter of 2026 as in 2024.
While one employee cited in the blog post said there are days when everything goes so well they “can’t help but think nothing I do matters,” another said it had been roughly five months since they last wrote any code themselves.
The report also said AI systems are increasingly capable of handling engineering and research work that once required humans.
While current models still struggle with higher-level judgment and deciding which problems are worth solving, the company said AI capabilities are advancing rapidly.
Anthropic stopped short of calling for an immediate pause.
However, it said that any meaningful slowdown would require coordination among multiple frontier AI developers and governments, warning that a unilateral pause by a single company would do little to improve safety.
While similar arrangements have existed for other powerful technologies, the company warned that building the infrastructure and trust required for them takes time.
“We don’t have that long,” it said.
Read the full article here
