Anthropic says it may soon reach its own warning line for AI that speeds up AI research

The San Francisco maker of Claude added the warning to its updated stock-sale filing Monday, while saying none of its current models has crossed the line.

Technology desk
Monday, Nov. 2, 2026In the Monday, Nov. 2, 2026 edition

Illustration of glowing orange arrows looping in a spiral past a row of circuit-board brains that grow larger and redder, ending in a huge blazing brain cracking its shell beside flashing red warning lights. — AI-generated illustration of a runaway feedback loop, in which AI helps build smarter AI, which then speeds up its own progress. Anthropic’s filing says none of its current models has crossed its warning line for that kind of acceleration. — AI-generated illustration

Anthropic, the San Francisco company behind the Claude chatbot, warned investors Monday that its AI may soon start speeding up AI research so much that it crosses one of the company's own safety lines. The warning appears in the risk section of an updated filing for its planned stock sale, submitted to the U.S. Securities and Exchange Commission.

our internal estimates and our expectations regarding future models, rather than our current models, indicate that we may imminently meet

The line it may meet is called "dramatically accelerated progress attributable to AI." Anthropic said it has not designated any of its current models as exceeding that threshold and is still studying the data.

What the warning means

Anthropic, like other AI labs, already uses its own AI to help write the code and run the experiments that build the next model. The filing says more than 90% of the contributions to its codebase in September were written by Claude.

The worry is a loop that some researchers call "recursive self-improvement": each model helps build a smarter one, faster and faster, until progress outruns people's ability to check that the systems are safe. The filing says Anthropic has in the past failed to notice a big speedup in its progress "until after it has occurred," and may again.

The company's safety rules

Anthropic's Responsible Scaling Policy is a set of rules it wrote for itself that says what extra safety steps it must take when its models reach certain abilities. The filing says Anthropic does not expect to be able to safely let Claude design and build new frontier models on its own without meaningful human supervision, and that it is considering updates to the policy "to keep humans in control."

TechCrunch, which first highlighted the language, noted that a stock-sale filing serves as both a legal disclosure and a sales pitch to investors. Anthropic said it will release further reports on the issue.

Source: Anthropic PBC amended Form S-1 (SEC, Nov. 2, 2026); TechCrunch.

Drafted by AI from the linked public source. Approved by a Daily News Editor.

Sources and documents