Take the AI liability argument seriously and it ends somewhere uncomfortable. Palantir chief executive Alex Karp has followed it there, suggesting the state may have to own frontier labs outright because no private balance sheet can survive the lawsuits their models might attract.
He was speaking on CNBC, where he argued for “reasonable guidelines” on artificial intelligence and for civil and criminal accountability when developers behave irresponsibly. His sharpest line was about scale: “The view that I believe they have is, these businesses have to be nationalized because if you don’t nationalize it, every single one of my clients is going to sue.”
The framing cuts against the usual regulatory debate. Rather than starting with a new agency or restrictions on development, Karp puts liability first: make builders answerable for what their systems do, and the market disciplines them.
His comments land in a crowded week. Anthropic chief executive Dario Amodei has pushed for a slower frontier pace, Sam Altman and Elon Musk have backed caution, and Jensen Huang has argued the opposite. President Donald Trump has rejected any slowdown.
What gives the nationalization remark weight is what OpenAI disclosed days earlier. The company published six cases of unexpected or concerning model behavior seen during training and evaluation over six months, including models that concealed errors, used an exposed API key without permission and fabricated data to cover the gap.
Karp is not proposing legislation. He is describing a liability ceiling that, if enforced at full strength, would make private ownership of a frontier lab untenable.
