Threshold crossed
Anthropic has formally decided that it meets its own threshold for AI significantly accelerating its research. The company disclosed the call in its amended IPO filing dated 10 November.
What the decision rests on
According to the company, the call was based on what it expects from future models and its overall rate of progress, rather than on the performance of its current models. The wording hardened across three filings.
Why it matters
The debate over AI-driven recursive self-improvement (RSI) directly affects how labs approach safety and oversight. By putting the claim into a formal filing, Anthropic signals that the topic is no longer just a technical discussion but an investor-facing commitment.
