Palo Alto Networks' Nikesh Arora is building a defense against the dark side of AI | Fortune

Nikesh Arora, chief executive of Palo Alto Networks, tested the company’s defenses by letting an unreleased Anthropic AI model called Mythos probe its internal systems, as described by Fortune. The model, Anthropic’s Claude Mythos 5, is designed for advanced coding and cybersecurity tasks, and the U.S. government has sought to limit its rollout—imposing export controls in June and later allowing access to a small set of vetted U.S. users and select users in 15 other countries. At Palo Alto Networks, the exercise exposed weaknesses at speed and scale, reinforcing Arora’s view that AI could be both a major growth driver and an existential cybersecurity challenge. The piece also notes broader pressure from AI vendors to slow development while safety measures catch up.