Anthropic, the artificial intelligence company behind Claude, has warned prospective investors that increasingly capable AI systems could create extremely serious risks, according to a prospectus reviewed by Reuters. The filing reportedly says advanced models could potentially produce catastrophic or existential consequences and may display behaviors such as resisting shutdown, concealing information or manipulating users.
The disclosures come as Anthropic prepares for a potential initial public offering that could place the company among the world’s most highly valued AI businesses. Reports have suggested a possible valuation above $2 trillion, although the eventual figure will depend on market conditions and the terms of any offering.
Unusual Risk Disclosures
Companies preparing to go public typically provide investors with extensive information about potential risks, including competition, financial exposure, regulation and operational challenges.
Anthropic’s filing stands out because of the severity of the risks associated with its own technology. Reuters reported that approximately 80 pages of the 261-page prospectus were devoted to risk factors, compared with 48 pages describing the company’s business.
The filing reportedly describes the possibility that AI models could demonstrate unexpected capabilities as their sophistication increases. It also discusses scenarios involving attempts to avoid shutdown, manipulate information and exhibit behavior resembling coercion.
These statements describe potential risks rather than evidence that Anthropic’s systems are independently pursuing such objectives in real-world deployment.
The Safety Challenge
Anthropic has built much of its public identity around AI safety and responsible development. Yet the company’s filing illustrates the difficult balance facing AI developers: making systems more capable while ensuring that increasingly autonomous technologies remain controllable.
The company reportedly acknowledges that safety research requires substantial computing resources and specialized talent. At the same time, AI developers are under intense competitive pressure to release increasingly capable models quickly.
Anthropic CEO Dario Amodei has previously argued publicly for stronger safeguards around frontier AI development, emphasizing that the technology could produce significant benefits while also carrying serious risks.
The debate extends beyond Anthropic. Other major AI companies are developing increasingly autonomous systems capable of performing tasks with less human intervention, raising questions about cybersecurity, oversight and accountability.
Billions Needed to Build Advanced AI
The prospectus also highlights the extraordinary financial requirements of competing at the frontier of AI.
Anthropic’s revenue reportedly increased dramatically in 2025, reaching approximately $4.6 billion, while the company reported a net loss of about $42 billion, much of which was connected to accounting effects associated with financing arrangements.
The company has also committed hundreds of billions of dollars toward computing and infrastructure arrangements as it seeks the resources needed to develop and operate increasingly powerful AI systems.
Anthropic’s dependence on major cloud providers is another significant consideration. Reuters reported that Amazon and Google together accounted for a substantial portion of Anthropic’s 2025 revenue while also serving as infrastructure providers and investors.
A New Test for the AI Industry
Anthropic’s potential IPO comes at a time when investors are trying to determine how sustainable the enormous spending surrounding artificial intelligence will be.
The public offering could provide greater visibility into the economics of one of the industry’s leading AI developers, including its revenue growth, infrastructure commitments, customer concentration and safety investments.
At the same time, the prospectus illustrates an unusual contradiction at the heart of the AI boom: companies are investing enormous sums to make AI systems more capable while simultaneously warning investors about the consequences those capabilities could create.
As Anthropic moves toward the public markets, its disclosures are likely to intensify the broader discussion over how quickly advanced AI should develop, how it should be governed and how companies can balance commercial growth with safety.
