Anthropic indicates that artificial intelligence is evolving at such a rapid rate that leading developers might have to reduce their pace.

Researchers from the company's dedicated research division explained in a Thursday blog entry that AI tools are already accelerating the creation of successor systems to an extent that could allow models to contribute directly to building improved versions of themselves.

"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 firm stated.

The statement arrives while AI transforms both the technology sector and wider economy at an unprecedented pace.

The timing coincides with Anthropic submitting confidential IPO registration documents, positioning the company ahead in the race with rival OpenAI to complete a public offering. OpenAI has yet to submit its own formal filings.

Over recent months leaders have highlighted AI's expanding capacity to execute tasks previously performed by people, including software coding, recruitment processes, and customer support operations. Google reported that AI now produces 75 percent of its code, and Mercor CEO Brendan Foody noted that his firm allocates more resources to AI usage than to staff compensation.

Meanwhile an increasing number of organizations have attributed workforce reductions and reorganizations to efficiency improvements driven by AI.

Anthropic shared internal metrics demonstrating the extent of task delegation to AI systems.

The data revealed that over 80 percent of code integrated into its repositories is generated by Claude, with average engineers merging eight times more code daily during the second quarter of 2026 compared with 2024.

One staff member quoted in the post described periods when results flow so smoothly that they "can't help but think nothing I do matters," while another indicated roughly five months had passed since they personally authored any code.

The analysis further noted that AI tools are taking on engineering and research responsibilities that historically demanded human expertise.

Although present systems still face challenges with advanced decision-making and identifying high-value problems, the company observed that capabilities continue to improve swiftly.

Anthropic refrained from recommending an immediate halt.

It emphasized that any effective slowdown would necessitate agreement among several leading AI organizations and regulatory bodies, cautioning that action by one firm alone would offer minimal safety benefits.

While precedents exist for managing other high-impact technologies through similar agreements, the post stressed that establishing the necessary frameworks and mutual confidence requires substantial time.

"We don't have that long," the statement concluded.

The publication drew pushback from segments of the technology community. Investor David Sacks responded on X by writing: "In other words, you want the government to save us from … you."