Article 293: AI Labs' Potential Liabilities Dwarf What Insurers Will Cover

The Wall Street Journal reports that specialist AI insurers estimate the largest labs' potential legal liabilities from safety incidents at "in the tens, maybe hundreds, of billions of dollars." Available AI insurance typically covers limits in the tens of millions. Some industry figures reportedly want Washington to stand behind the losses.

Key Takeaway: Insurance is the quietest form of regulation there is, because an underwriter's refusal is a decision that requires no statute and admits no appeal. Here the gap between the risk and the coverage is three or four orders of magnitude, which means the market has effectively declined to price AI safety incidents at all. That leaves two options, and the industry is asking for the second: carry the risk on the balance sheet, or transfer it to the public. A backstop would not make AI safer. It would make the losses survivable while removing the last financial incentive to prevent them - and it would be written by the same governments currently arguing that the sector should not slow down.

Sources: The Wall Street Journal, Business Model Analyst, 17 October 2026