Anthropic said in an amended prospectus that it has met a safety threshold under its Responsible Scaling Policy, based on expectations for future models and the company’s overall rate of progress rather than observed capabilities of current models. The threshold describes a scenario in which AI performs enough of Anthropic’s research to roughly double the pace of model improvement, compressing two years of progress at the company’s 2018-2024 rate into one year. Crossing it requires additional safeguards rather than a shutdown, and Anthropic said it is still assessing the consequences and plans to publish further reports. The company said AI has already accelerated its research and development, while warning that broader use of Claude could make some processes difficult for humans to understand and control. METR estimated that Anthropic’s models have been improving nearly twice as fast since October 2025 and said the company’s earlier claim that it remained below the threshold was overstated, with the threshold plausibly crossed by mid-July. Anthropic said it does not expect to use Claude to fully autonomously design and build successive generations of frontier models without meaningful human supervision or understanding, and is considering policy changes to keep humans in control. The disclosure came as an updated filing set the IPO range at $70 to $78 per share, valuing Anthropic at about $1 trillion on shares outstanding and up to $1.4 trillion on a fully diluted basis. The offering covers 825 million Class A shares and could raise up to about $64.4 billion, while a private placement of non-voting Class N shares seeks $26 billion. Google and Nvidia are each reportedly committing $10 billion, with Microsoft also involved, bringing potential total proceeds to around $90 billion. Anthropic reported $81 billion in run-rate revenue as of October 31, 2026. Pricing is expected November 18, with Nasdaq trading projected around November 19 under ticker ANT. The company’s Long-Term Benefit Trust is positioned to control a majority of the board, limiting public shareholders’ influence. Seven co-founders plan to sell approximately $1.75 billion of stock, while the filing also details AI-related insurance risks, lockups and terminology changes tied to President Trump’s Sept. 29 executive order. The offering will test whether Claude can develop into a broad software platform capable of supporting a valuation above $1 trillion, rather than remain a capital-intensive AI business.