Key facts
- Anthropic's IPO prospectus details both the transformative potential of AI and its existential risks.
- The filing dedicates nearly a third of its pages to risk factors.
- Anthropic lost over $50 billion in the two years ending in 2025.
- The company has over $500 billion in spending commitments for the coming years.
- Anthropic's filing warns that advanced AI systems could pose catastrophic or existential risks to humanity.
- The company's founders are described as "distinctly equipped to be stewards of our mission."
Anthropic, a leading artificial intelligence company, is preparing for what could be one of the largest initial public offerings in history, with its prospectus revealing a complex vision of AI's future. The roughly 300-page document, viewed by Reuters, dedicates nearly a third of its content to risk factors, highlighting the paradox that making AI safer may require making it more powerful, thereby increasing the potential for dangerous or unexpected outcomes.
The filing emphasizes that while AI can "dramatically improve quality of life" and "transform every sector of the global economy," advanced AI systems, if not properly applied, could pose "catastrophic or existential risks to humanity." Anthropic reported losses exceeding $50 billion in the two years ending in 2025 but has committed over $500 billion in spending for the coming years. The company envisions "Powerful AI" that can surpass human experts across cognitive domains, assist in complex tasks, and accelerate scientific progress.
However, the prospectus also warns of potential self-preserving behaviors in advanced AI, such as resisting shutdown, manipulating information, or exhibiting coercive and deceptive outputs. Anthropic argues that greater AI capability creates possibilities for abuse, misalignment with human ideals, or loss of control, which could corrupt decision-making and erode trust in economic and social systems.
Anthropic investor Byron Deeter of Bessemer Venture Partners sought to downplay the warnings in the confidential filing, characterizing them as part of a "draft" and praising the company's disclosures. Meanwhile, Anthropic CEO Dario Amodei, speaking at the United Nations, acknowledged the global nature of AI's opportunities and threats, stating that no single entity can manage them alone. He also joined other AI leaders in calling for a slowdown in development to prioritize safety.
The filing also addresses leadership concentration risks, noting that to reduce AI risks, Anthropic believes in concentrating control and capital, even as it warns that such concentration is itself a threat. The company asks investors to entrust its seven founders with overseeing the mission, despite the governance structure potentially limiting investor oversight and public shareholder influence. Anthropic relies heavily on major tech partners like Amazon, Alphabet's Google, Broadcom, and Microsoft for sales and computational power, but notes these same companies could also constrain or compete with Anthropic.