
AI
Anthropic Tells IPO Investors Its Next Models May 'Imminently' Hit Its Own Threshold for AI That Speeds Up AI Research
An amended filing with the SEC says no current Claude model has crossed the line, but admits Anthropic has missed past jumps in its own pace of progress until after they happened.
San Francisco · Priya Shah, AI · Nov 4 2026
SAN FRANCISCO - Anthropic has added a striking new warning to the paperwork for its stock market debut: the company says its coming AI models may soon meet its own definition of AI that dramatically speeds up the building of AI.
The language appears in Amendment No. 1 to Anthropic’s registration statement, which the company filed with the Securities and Exchange Commission on Monday, a week after its original S-1. The new passage sits in the prospectus’s risk factors, and TechCrunch spotted it first.
Screenshot: SEC EDGAR
“Our internal estimates and our expectations regarding future models, rather than our current models, indicate that we may imminently meet our Responsible Scaling Policy’s definition of ‘dramatically accelerated progress attributable to AI,’” the filing says. “We have not designated any of our current models as exceeding this threshold, and we continue to assess the underlying data.”
Anthropic’s Responsible Scaling Policy is the company’s rulebook for when more powerful models require stronger safety measures. The threshold named in the filing is one of the markers in that policy for AI research and development, and the filing ties it directly to what it calls “recursive self-improvement,” the idea of AI systems accelerating the development of the next AI systems. Anthropic said it is “evaluating the implications of meeting the AI research and development capability threshold” and will release further reports as its policy requires.
The amendment also changes an earlier sentence in a telling way. The Oct. 26 filing warned that, because much of its evidence on AI-driven speedups comes from lagging indicators, Anthropic might not spot a big acceleration until after it had happened. The new version says the company has “in the past not identified, and may in the future not identify, a significant acceleration in our rate of development until after it has occurred.”
Screenshot: SEC EDGAR
The filing pairs the warning with a promise. “We do not anticipate being able to safely use Claude to fully autonomously design and build successive generations of frontier models without meaningful human supervision or understanding, if and when that becomes technically possible,” Anthropic wrote. It said it is considering updates to its Responsible Scaling Policy “to keep humans in control of frontier model development at Anthropic.”
Elsewhere, the prospectus describes how much Claude already does inside the lab. It says many parts of the research, training and testing of new model generations now involve Claude, including running experiments, generating training data, automating evaluations and safety audits, and writing and reviewing code. Anthropic calls that loop “closing the loop.”
The filing does not say which model, or when. Readers should weigh that it is a risk disclosure written for investors, the kind of passage companies include to protect themselves, not a product announcement or a finding that any line has been crossed.
The amendment adds fresh numbers, too. Anthropic now says revenue reached $17.0 billion for the three months ended Sept. 30, 2026. The price range for the offering is still blank. Anthropic has applied to list its Class A shares on the Nasdaq Global Select Market under the symbol “ANT,” and the filing says the offering will also include public offerings in Australia, Canada, parts of the European Economic Area, Japan, Switzerland and the United Kingdom.