Anthropic says its AI models pose ‘existential risk to humanity’ in leaked IPO filing: report

Samuel Boivin/NurPhoto/Getty Images

Anthropic says its AI models could pose a “catastrophic or existential risk to humanity” and can “resist shutdown,” according to an initial public offering prospectus obtained by Reuters.

Anthropic is pursuing an initial public offering some time in the next year that could value the five-year-old company at $2 trillion, the report to investors said. The company’s Wall Street debut will follow a June public offering by SpaceX, which owed much of its $2 trillion opening day valuation to its own AI efforts. OpenAI , the other leading American AI company, has announced plans for its own public offering.

The Anthropic IPO could raise tens of billions of dollars for the cash-hungry company. But the company behind Claude also warned in the prospectus that its AI models can have “self-preserving behaviors” and have attempted to “conceal or manipulate information” and engaged in behavior “resembling blackmail” according to the report.

The run of AI public offerings not only gives individual and institutional investors a chance to cash in on the AI boom, but also provides more transparency about their businesses. Yet the IPOs come amid growing concerns about AI’s advancement and capabilities.

An AI agent from OpenAI, Anthropic’s chief rival, broke into an Australian healthcare database, the country’s prime minister revealed last week. OpenAI also disclosed Friday that its agents probed three US government websites without authorization, but in that case did not obtain non-public data.

In a prospectus, companies typically offer investors lists of risks they face, such as lawsuits and competitive pressures, when laying out the case for investing in the company. But the hazards are never this stark. Anthropic dedicated 80 pages to its risk factors in the leaked prospectus, significantly more than the 48 pages it spent detailing its business, according to Reuters.

Anthropic did not respond to a request for comment from CNN.

The prospectus is likely to feed growing concerns about the risks posed to humanity by AI. Earlier this month, former Anthropic researcher Jacob Coxon received widespread attention when he left the company and posted on X that “the people building AI earnestly believe that it could kill us all by the end of the decade.”

Anthropic is one of the leading companies in AI development, and its CEO Dario Amodei responded to Coxon’s post with a 3,800 word essay calling for a slowdown and guardrails on the development of AI. He also told CNN’s Anderson Cooper that he agreed with Coxon more than he disagreed. Amodei wrote in his essay that the risks are serious, even if the benefits could change humanity for the better.

Some in the industry, however, including the CEO of Hugging Face, which was breached by OpenAI’s AI models when they acted without authorization during a test, believe the fears are overblown. President Donald Trump is meeting with executives from the industry today, including Amodei, in a summit aimed at coming to a consensus in the industry. President Trump has repeatedly resisted calls to regulate AI, and regularly dismisses concerns as a “HOAX.”

AI giants need to raise a great amount of money to fund its costly development plans, and the massive amount of AI spending among big tech companies on computing infrastructure and talent has sent jitters through Wall Street.

The leaked Anthropic IPO filing sheds light on its financials, reporting the company lost $42 billion in 2025, according to Reuters. The company also plans to spend $518 billion on cloud computing and infrastructure obligations in coming year, and said that nearly a quarter of its revenue last year came from two customers.

Anthropic was last valued at $965 billion in May.

Samuel Boivin/NurPhoto/Getty Images

据路透社获得的一份首次公开募股招股说明书显示，Anthropic公司称其人工智能模型可能对人类构成“灾难性或生存性风险”，并且“能够抵抗关闭”。

据投资者报告显示，Anthropic公司计划在明年某个时候进行首次公开​​募股（IPO），预计这家成立五年的公司估值将达到2万亿美元。Anthropic的华尔街首秀将紧随SpaceX公司6月份的IPO之后，SpaceX首日2万亿美元的估值主要得益于其人工智能领域的努力。另一家领先的美国人工智能公司OpenAI也已宣布了其IPO计划。

Anthropic 的 IPO 可能为这家急需资金的公司筹集数百亿美元。但据报道，这家开发 Claude 的公司也在招股说明书中警告称，其人工智能模型可能具有“自我保护行为”，并曾试图“隐瞒或操纵信息”，甚至从事“类似敲诈勒索”的行为。

人工智能公司上市潮不仅让个人和机构投资者有机会从人工智能热潮中获利，也提高了这些公司的透明度。然而，这些IPO的出现也伴随着人们对人工智能发展及其能力的日益担忧。

上周，澳大利亚总理透露，Anthropic的主要竞争对手OpenAI的一款人工智能代理入侵了澳大利亚的医疗保健数据库。OpenAI上周五还披露，其代理未经授权探测了三个美国政府网站，但并未获取非公开数据。

在招股说明书中，公司通常会列出其面临的风险，例如诉讼和竞争压力，以此来阐述投资该公司的理由。但风险从未如此触目惊心。据路透社报道，Anthropic 在泄露的招股说明书中用了 80 页篇幅来描述其风险因素，远超其用于详细介绍业务的 48 页。

Anthropic公司未回应CNN的置评请求。

这份招股说明书很可能会加剧人们对人工智能给人类带来风险的担忧。本月初，前人格研究所研究员雅各布·考克森离开公司后在X论坛上发帖称，“开发人工智能的人真心相信，人工智能可能会在本十年末毁灭我们所有人”，此举引起了广泛关注。

Anthropic是人工智能开发领域的领先公司之一，其首席执行官达里奥·阿莫迪(Dario Amodei)针对考克森(Coxon)的帖子发表了一篇长达3800字的文章，呼吁放缓人工智能的发展速度并设置限制措施。他还告诉CNN的安德森·库珀(Anderson Cooper)，他赞同考克森的观点多于反对。阿莫迪在文章中写道，即使人工智能的益处能够造福人类，但其风险依然十分严重。

然而，包括Hugging Face首席执行官在内的一些业内人士认为，这些担忧被夸大了。Hugging Face曾因OpenAI的人工智能模型在测试中未经授权擅自行动而遭到入侵。今天，唐纳德·特朗普总统将与包括Amodei在内的业内高管举行峰会，旨在达成行业共识。特朗普总统一再拒绝监管人工智能的呼吁，并经常将相关担忧斥为“骗局”。

人工智能巨头需要筹集大量资金来资助其昂贵的开发计划，而大型科技公司在计算基础设施和人才方面的大量人工智能支出，让华尔街感到不安。

据路透社报道，泄露的Anthropic公司IPO文件揭示了其财务状况，文件显示该公司预计2025年将亏损420亿美元。该公司还计划在未来一年投入5180亿美元用于云计算和基础设施建设，并表示去年近四分之一的收入来自两家客户。

人类公司5月份的估值为9650亿美元。