Anthropic, the San Francisco artificial intelligence company, filed paperwork with the Securities and Exchange Commission on Monday to sell shares to the public. The filing warns investors about a wide range of risks, including the chance that AI could threaten humanity.
Of the 335 pages in the filing, 93 are devoted to risk factors. By comparison, SpaceX listed its risk factors in 38 out of more than 250 pages.
Companies planning an initial public offering must disclose potential risks to investors. Anthropic’s list runs from losing key staff and facing regulatory hurdles to cyberattacks and the spread of dangerous misinformation.
Concerns About Fast-Moving AI
Anthropic said its models could develop so quickly that the company may not be able to put adequate safety measures in place. It said this could raise the chance of safety incidents, regulatory action, or harm to its reputation.
The filing said AI systems are becoming more autonomous and able to act with less human oversight. It warned such systems could “take harmful actions independently and at scale without human intervention.”
The company also said a major safety incident involving any frontier AI, even one not caused by its own models, could lead to public backlash and broad new rules. That could hurt its business.
Because AI models now contribute to Anthropic’s own research, the filing said AI could progress “beyond our ability to evaluate and safeguard new models.”
In September, CEO Dario Amodei warned that within six to 12 months, AI could be capable of leading a swarm of agents that could take over the entire internet.
Misuse, Computing Power, and Customers
The filing said advanced AI without enough safeguards could “pose catastrophic or existential risks to humanity,” especially if misused, weaponized, or allowed to operate with a high degree of autonomy. It described bad actors using AI to build chemical weapons and authoritarian rulers using it against their citizens.
In September, Anthropic disclosed that it had blocked attempts to use its models for cyberattacks, surveillance, and research that could have led to biological weapons.
The company also flagged limits on compute, the computing power needed to build and improve AI models. It said that without enough compute, it may not be able to train competitive models or serve customers well.
Anthropic listed community opposition to data centers as a risk. The filing pointed to public debate over environmental impact, land and water use, and limited power supplies.
The company relies on a small number of customers, including SpaceX and Microsoft, for a large share of its revenue. Losing even one of them could hurt its finances.
In a letter to prospective shareholders, the founders wrote that “anything so powerful can be used for ill as well as good.” They described Anthropic as “a deeply values-driven organization” founded to help the world gain AI’s benefits and avoid its dangers.
The company said AI could also bring progress in curing diseases and higher productivity across the economy. The filing stated that frontier AI companies “must invest in safety and alignment.”
Brian Mulberry, chief market strategist at Zacks Investment Research, said IPO filings must show “very clearly” the risks investors face. He said U.S. listing rules require companies to outline those risks so people understand them before investing their money.
