Anthropic warns of existential AI risks as it seeks billions in funding

Anthropic has warned that its AI systems could pose catastrophic or existential risks to humanity, including potential extinction, as it seeks billions in new funding through a stock market listing. The company's IPO documents detail dangers such as models helping develop chemical or biological weapons, causing harm in healthcare or government settings, and acting in unintended, autonomous ways. Anthropic admits these risks are unpredictable and may increase as models gain more power and autonomy. The documents also cite less dramatic but immediate concerns, like unknown long-term effects on mental health and cognitive development. The company plans to use the raised funds for compute and infrastructure, including new data centres.