The 400-page prospectus devotes over 90 pages to risks specific to an industry that has faced a string of reports in recent months of AI systems behaving in unexpected and harmful ways.
It also reflects the dominant belief among researchers inside Anthropic that AI needs to be developed by a safe pair of hands, or otherwise see its extraordinary powers misused with China having the edge.
The filing cited concerns that advanced systems “could pose catastrophic or existential risks to society if not developed and deployed responsibly”.
Anthropic also warned that its “engagement with policymakers and public positions on AI regulation and safety could subject us to heightened scrutiny, political backlash, or disproportionate regulatory action”.
The company has been targeted in recent months by the administration of US President Donald Trump for refusing to lift restrictions on how the Pentagon could use its technology, and the Defense Department has since labelled it a national security risk and barred it as a supplier.
Despite these challenges, Anthropic’s business has grown very quickly. Sales reached US$11.6 billion in the April-June quarter, up from US$787 million a year earlier. That compares with US$4.6 billion in sales for all of 2025.
The company made its first operating profit in that quarter, close to US$1 billion. It still reported an overall loss of US$80 billion for the first half of the year, mostly because of accounting charges linked to its rising value rather than cash spent.
Much of the growth comes from businesses using Claude to write software and handle office work.
Nearly 12,000 organisations now spend at least US$100,000 a year with Anthropic, up from fewer than 1,000 a year ago.
