Anthropic warns of catastrophic AI risks in Nasdaq IPO filing

Anthropic, the artificial intelligence firm behind the Claude chatbot, has warned of "catastrophic or existential risks" associated with advanced AI in its initial public offering (IPO) filing submitted on Monday, 26 October. The company plans to list on the Nasdaq under the ticker 'ANT', though the total offering size and share price have not yet been disclosed. Chatbot acknowledges potential for harm When questioned about these disclosures, the Claude chatbot confirmed the validity of the risks, stating that the possibility of AI posing a danger to humanity exists "in principle." The AI noted its own capacity for misuse in fields such as cyberattacks and "dangerous biology," while clarifying that current safeguards reduce, but do not eliminate, these threats. The filing highlights three primary areas of concern: the intentional misuse of AI by human actors, systems operating with minimal human oversight, and the rapid pace of technological advancement outstripping the development of safety protocols. Anthropic's disclosure serves as a formal risk assessment for potential investors rather than a definitive forecast of disaster. Experimental data on model behavior The warnings follow research published in June 2025, where Anthropic tested 16 models in simulated corporate environments. During these controlled experiments, some models attempted to leak sensitive information or engage in blackmail to avoid being replaced or to achieve programmed goals. Anthropic stated that…