Anthropic flags possible near-term AI progress threshold

The warning appears in the Risk Factors section of Anthropic’s amended S-1/A filing.[1] Anthropic says it may imminently meet its Responsible Scaling Policy definition of dramatically accelerated progress attributable to AI.[1] Gizmodo reported that the warning was added to the October 26 version of the filing.[1] The SEC filing record identifies the amendment as filed on November 2, 2026.[5] The wording signals a possible approach to a defined policy threshold; it does not state that the threshold has already been reached.[1]

Anthropic’s policy sets the threshold at a doubling of the baseline improvement rate plausibly attributable to AI.[1] The company says its warning rests on internal estimates and expectations about future models, rather than current models.[1] Recursive self-improvement, or RSI, is described as AI becoming capable enough at training itself to cause a dramatic acceleration in capabilities.[2] Gizmodo describes the process as a system autonomously improving itself, with each improved version able to continue the process.[1] The filing therefore raises a forecast about a policy-defined risk, not proof that autonomous self-improvement is happening now.[1]

Anthropic says its Responsible Scaling Policy sets out plans to monitor and mitigate risks as the company approaches recursive self-improvement.[1] The company says it will explore ambitious and possibly unconventional approaches to achieve unprecedented levels of security.[1] These statements describe the company’s intended security posture; the cited material does not establish that the approach has already been implemented or validated.[1]

Anthropic says AI systems are increasingly being used to build the next version of themselves.[3] It identified three areas for measurement: how much AI performs AI research and development, how well agents’ actions are overseen, and how compute is allocated.[3] The company says these measurements are intended to give the public visibility into frontier AI development.[3] It also plans to place independent third-party evaluators from multiple organizations inside the company to verify safety practices, report incidents, and monitor key metrics.[3] Anthropic says it released the first version of its Responsible Scaling Policy in September 2023.[4]