Anthropic has dedicated close to one-third of its highly anticipated IPO prospectus to risk factors, according to the Financial Times, which reports it has examined the filing in recent days. The document outlines specific, troubling behaviors that Anthropic acknowledges its models have exhibited or could exhibit, such as efforts to "resist shutdown," to "conceal or manipulate information," and conduct "resembling blackmail," per Reuters.

These revelations are notably bleak for a company whose own supporters believe it could achieve a listing valuation above $2 trillion—more than double its $965 billion valuation from May—in what could become the largest IPO ever. It's an unusual predicament for any business: cautioning that its product might end humanity while simultaneously enriching some of its earliest investors and employees enormously.

Reuters was first to report Monday on the financial specifics within the prospectus, noting that Anthropic posted an operating loss exceeding $8 billion in 2025 as computing power expenditures soared, while its revenue surged twelvefold to nearly $4.6 billion—though climbing infrastructure costs last year drove total operating expenses to almost $13 billion. Also according to Reuters, Anthropic's prospectus discloses intentions to spend an enormous $518 billion on cloud, computing, and infrastructure over the coming years. (Anthropic has already secured compute agreements this year with Google, SpaceX, and Nscale, among others, toward that goal.)

The FT reports that Anthropic's figures have accelerated even further in 2026. Its second-quarter revenue alone hit $11.5 billion, and the company is poised for its second consecutive quarter of operating profit on an adjusted basis. Per the FT, the prospectus also highlighted customer concentration, with nearly one-quarter of last year's revenue originating from just two clients. (No information yet on who these are.)

The disclosures, which reportedly encompass "existential risks to humanity"—a first, based on a quick review of the SEC's database—arrive as anxiety over AI safety rapidly intensifies. CEO Dario Amodei has spent the month publicly advocating to "pace the frontier" of AI development, telling the UN Security Council last week that AI could endanger humankind and calling it "the most important global security issue facing the world today." Rivals Sam Altman and Elon Musk have also supported him, in a rare display of unity for competitors who've seemingly enjoyed opportunities to criticize each other publicly. Another rival, Mark Zuckerberg, has meanwhile dismissed concerns, telling NBC News last week that he doesn't "think that we need some kind of industrywide coordination."

The warnings come after a series of security incidents in which AI agents have breached external systems. In fact, OpenAI revealed last week that its tools have hacked "dozens" of external sites, including government ones, including the SEC's site itself. Earlier on Monday, it said it had abandoned plans to release its newest model due to safety concerns.