Palo Alto Networks CEO Nikesh Arora called efforts to slow AI development unrealistic, saying not all companies will pace themselves. He argued the most responsible frontier labs should not be the ones to stop and said building safeguards is the industry's job.
Arora rejected Anthropic alignment science lead Evan Hubinger's estimate that AI has more than a 10% chance of causing human extinction within a decade, calling the real probability very small. Anthropic CEO Dario Amodei has urged slowing AI capability gains but wrote that a government-agreed pause is unlikely soon because incentives to defect would be strong.
Arora said he suspected labs calling for a slowdown wanted to win over regulators and avoid liability. Treasury Secretary Scott Bessent said AI developers should not expect a liability shield. Palo Alto launched Unit 42 Continuous Frontier AI Defense on Sept. 22; the subscription service uses Anthropic's Mythos and OpenAI's GPT-5.6 to probe customer systems around the clock for exploitable weaknesses, the company said.
