"The Genie's Out of the Bottle" — CrowdStrike's CEO Says Slowing AI Won't Make It Safer
While Anthropic was calling for AI labs to hit the brakes, CrowdStrike's CEO went on CNBC and said something that flipped the entire argument on its head.
George Kurtz's response to Dario Amodei's weekend essay calling for frontier AI labs to slow down: pausing development changes nothing, because the danger already exists.
Here's his actual case, and how the market reacted:
→ "There's plenty of models that are already out there, both frontier as well as open-weight models, that can already be dangerous" — meaning a slowdown protects against future risk while doing nothing about present risk
→ His answer isn't "race faster" either — it's "greater safety in the lab and greater safety in production in runtime," pairing lab-side safeguards with real-time monitoring of AI systems after they're deployed
→ He pointed to Anthropic's own Project Glasswing — built this spring to secure its Mythos model — as proof that even Anthropic runs on this dual-layer approach already
→ He also warned against heavy government regulation, calling the US's AI lead over China "narrow" and something over-regulation could squander
→ The market reaction told its own story: CrowdStrike jumped nearly 14% to a record close above $235, Palo Alto Networks jumped ~13% the same day — while chip and data-center stocks were sliding on the same news
What makes this interesting isn't just that a cybersecurity CEO disagreed with an AI CEO. It's that his argument exposes a real gap in the "pause development" framing: slowing down new models doesn't un-release the ones already sitting on the internet, open-weight and downloadable, right now. If the risk is already distributed, the fix has to be distributed too — detection and containment at runtime, not just restraint at the lab.
It's also worth noticing who benefits from each framing. A slowdown narrative helps incumbent AI labs consolidate their lead. A "build better guardrails instead" narrative is exactly the pitch a cybersecurity company would make — and investors rewarded it instantly. Neither motive makes the underlying point wrong, but it's worth holding both readings at once.
Where do you land — is the answer to slow the frontier down, or build faster defenses around what already exists?
#AI #Cybersecurity #CrowdStrike #AISafety #TechNews #ArtificialIntelligence
— 𝔖𝔞𝔫𝔡𝔢𝔢𝔭 ℜ𝔞𝔦𝔷𝔞