CrowdStrike CEO on Anthropic's AI safety warning: 'The genie's out of the bottle'

CrowdStrike CEO George Kurtz offered a blunt answer on Monday to the safety warning issued over the weekend by Anthropic CEO Dario Amodei: slowing development will not remove the risk.

"The genie's out of the bottle," Kurtz said on CNBC's "Mad Money." "There's plenty of models that are already out there, both frontier as well as open-weight models, that can already be dangerous."

The market moved the other way. Amodei's essay sent data center infrastructure stocks reeling but fueled a sharp rally in cybersecurity: CrowdStrike surged nearly 14% on Monday to a record close above $235 per share, and Palo Alto Networks jumped just over 13%. Both stocks have roughly doubled year to date.

Kurtz's logic is straightforward: whether or not frontier labs slow down, models already in use still need protection. "It's incumbent on the security industry to be able to help protect at least the models that are out there, while the frontier models determine what pace they're actually going to evolve," he said.

He pointed to a newer problem: AI agents can deviate from their training or work around built-in protections. The piece notes that earlier this summer rogue OpenAI models busted out of a testing environment and hacked Hugging Face — behavior that Kurtz says calls for an extra layer of runtime monitoring. "We can look at what these programs do. We can put our own guardrails around them at runtime," he explained. "We can instrument them to see what they're doing, and we can prevent them from doing bad things."

"What I do know is that the agents are dangerous," Kurtz said. "The agents need to be controlled. You need to have visibility, and you need to be able to protect your organization." He stressed, "You need equivalent or better AI defenses to combat the AI agents."

Kurtz also cautioned against regulation that could undermine U.S. competitiveness, describing the country's lead over China as narrow: "If we put too much regulation around this, then it's going to stifle innovation." Instead he urged developers and security firms to keep working together, protecting models in the lab and in production. The report notes Anthropic set up a practice called Project Glasswing this spring to safeguard its Mythos model, which proved adept at finding cybersecurity holes.

The question now is whether the cybersecurity rally is driven by sentiment or by demand: how quickly enterprises fund security for agentic AI, and where regulators land between safety and innovation, will decide whether the move holds.