Anthropic IPO filing warns of rogue AI, cyberattacks, and weapon risks

Anthropic has disclosed risks including rogue AI, cyberattacks, and weapon development in its IPO filing ahead of its stock market debut. The 335-page document devotes 93 pages to risk factors, far exceeding rival SpaceX's 38 pages. The company warns that rapidly developing AI models could outpace safety measures, potentially leading to harmful autonomous actions or regulatory intervention. It also cites risks from malicious actors misusing technology for chemical weapons or surveillance, alongside operational concerns about compute shortages and customer concentration. Anthropic frames its IPO as a step toward mitigating AI dangers, positioning itself as a responsible AI company. The filing notes that any significant safety incident in the industry could trigger reputational damage and sweeping regulatory responses affecting its business.