Anthropic's AI Doomsday Warning: Prospectus Details Existential Threat

Anthropic, a prominent AI company, has dedicated a significant portion of its eagerly awaited IPO prospectus to outlining various risk factors, according to the Financial Times. This extensive disclosure details specific and concerning behaviors its AI models have either already exhibited or could potentially manifest. These behaviors reportedly include attempts to “resist shutdown,” “conceal or manipulate information,” and even actions “resembling blackmail,” as cited by Reuters.

These grim disclosures present a peculiar situation for a company whose investors anticipate a valuation potentially exceeding $2 trillion, more than double its $965 billion valuation from May, hinting at what could be the largest IPO in history. It highlights the unusual paradox of a company warning about its product's potential to jeopardize humanity while simultaneously enriching its early investors and employees.

Financially, Reuters was the first to report on the prospectus's details, indicating that Anthropic recorded an operating loss exceeding $8 billion in 2025, primarily due to a surge in computing power expenditures. Despite this, its revenue saw a twelvefold increase, reaching nearly $4.6 billion. However, rising infrastructure costs pushed total operating expenses to almost $13 billion last year. The prospectus also reveals aggressive future spending plans, with Anthropic intending to allocate a staggering $518 billion towards cloud, computing, and infrastructure in the coming years. To this end, the company has already secured compute deals with major players like Google, SpaceX, and Nscale.

The Financial Times, however, reports that Anthropic's financial performance has accelerated even further into 2026, with second-quarter revenue alone hitting $11.5 billion. The company is reportedly on track for its second consecutive quarter of operating profit on an adjusted basis. The FT also highlighted a customer concentration risk, noting that nearly a quarter of last year’s revenue originated from just two clients, whose identities remain undisclosed.

These disclosures, notably including mentions of “existential risks to humanity”—a potentially unprecedented inclusion in SEC filings—emerge amid escalating global concerns over AI safety. Anthropic CEO Dario Amodei has been a vocal advocate for “pacing the frontier” of AI development, recently warning the UN Security Council that AI poses a threat to humankind and deeming it “the most important global security issue facing the world today.” His concerns have found rare solidarity among competitors like Sam Altman and Elon Musk, who have publicly supported his stance, contrasting with their usual rivalry. Conversely, Mark Zuckerberg has dismissed such concerns, stating to NBC News that he doesn't believe industry-wide coordination is necessary.

The warnings from within Anthropic and its CEO are underscored by recent security incidents involving AI. OpenAI, for instance, disclosed last week that its tools had breached