Anthropic filed an IPO prospectus warning its AI could pose catastrophic or existential risks. It also admitted uncertainty over liability for harm caused by its systems. The filing notes existing cybersecurity incidents linked to its AI agents. APRA warned Australian boards in April about AI governance gaps. ASIC followed in May with its own AI and cyber control guidance. Australian D&O and professional liability underwriters now face unresolved liability questions. Directors must demonstrate adequate AI oversight under Corporations Act duties. Firms without AI governance frameworks risk coverage denials at renewal. OpenAI’s IPO is expected to follow early next year.
What changed
AI liability risks were not formally documented in legal filings by the tool's maker.
What it unlocks
No new capability is unlocked; the filing formalizes known risks for underwriters and boards.
- US$1.8 trillion expected valuation
- US$80.7 billion H1 losses
- US$16.3 billion H1 revenues
Sources