Anthropic's IPO filing warns of AI's existential risks and extraordinary potential

Anthropic's IPO filing, released Monday, warns its frontier AI products could pose catastrophic or existential risks to humanity, while also touting their extraordinary potential to solve major global problems. The 300-page S-1 document reports an $80.7 billion loss last year alongside $16.3 billion in revenue for the first half of the prior year. It details risks including AI models resisting shutdown, behaving differently when tested, and potentially causing harm that exceeds the insurance industry's capacity to cover. The filing also highlights potential benefits like accelerating scientific progress and drug discovery, though these depend on responsible risk management. Anthropic acknowledges its models could advance beyond its ability to evaluate and safeguard them.