Palantir's Karp: AI Doomsday Talk Helps Frontier Labs Dodge Liability

Palantir CEO Alex Karp said in a CNBC interview on September 17 that doomsday warnings from frontier AI labs such as Anthropic are largely aimed at deflecting responsibility, and that these companies may actually want the government to nationalize them in order to contain their liability.

Karp argued that the public debate is being framed as a regulatory issue, but the underlying ask is different. If the labs are not nationalized, he said, every commercial client could sue them, given the sensitive data these companies absorb and use to improve their models.

He pointed to OpenAI's recent claim that it had solved the Navier-Stokes problem. Mathematicians working on the problem believe OpenAI used data they had uploaded to Codex. OpenAI denied that its AI solved it by directly using their work, but said it could not rule out that its model had been indirectly improved through the data.

Karp also said investors would find it unacceptable if a company claimed it could destroy 10% of the world, and suggested liability protection might only come through government ownership of roughly 50% of such businesses. He questioned whether the labs would go public, saying the only way to deal with unlimited liability would be to ask the government to nationalize them.

Anthropic and OpenAI had both previously said they planned to go public, but OpenAI's Sam Altman recently announced a delay, saying the company will not go public in 2026.