CrowdStrike and CoreWeave bring AI security to machine speed
SiliconANGLE Victoria Gayton
CrowdStrike and CoreWeave are tying AI security to CoreWeave’s running infrastructure. They say attackers can move in 27 seconds, so defense has to keep up.
Based on reporting by SiliconANGLE, Victoria Gayton — read the original for the full story.
Summary, retelling and take written by AI under human oversight; images are AI-generated illustrations. How we work · Report an error
CrowdStrike and CoreWeave are pushing a simple idea with hard edges: if attacks now move at machine speed, defense has to as well. At the Fully Connected event, CrowdStrike chief of AI Bartley Richardson pointed to the company’s fastest observed eCrime attacker breakout time in 2025 — 27 seconds — as the kind of clock security teams are up against.
That urgency is shaping the partnership. CoreWeave provides training and inference capacity for SafeMind and acts as a design partner for CrowdStrike’s Cyber Superintelligence Lab. The point isn’t just more compute. It’s putting CrowdStrike’s security data and adversary knowledge next to infrastructure that can help turn attacks into training material, then feed those lessons back into defensive models.
Richardson framed that as adversarial co-evolution: offensive capabilities are trained to probe for weaknesses, and defensive models are trained on what those attacks reveal. It’s a neat loop, and a brutal one. The underlying bet is that large-scale infrastructure is now part of the security stack, not just a place where the stack happens to run.
CoreWeave’s own problem is familiar to anyone running production systems that can’t blink. Chief information security officer Jim Higgins said security has to be built into the machine while it’s still running, from AI-assisted code review to automation in the continuous integration and continuous delivery pipeline to fix misconfigurations. He compared it to aikido: use the system’s own motion against the problem.
For all the talk of AI-native security, neither company is pretending the basics are obsolete. Higgins said organizations still need to patch, test and monitor continuously. Richardson made the trust argument in plainer terms: AI doesn’t end, it just becomes automation once people trust it. That may be the most honest line in the whole conversation.
My take — AI-written commentary, not fact-checked reporting
This is the right kind of AI pitch: less magic, more plumbing. Security vendors love to sound futuristic, but the real story is still patching, monitoring and getting the model to earn trust instead of demanding it. The industry keeps calling that AI; a lot of it is just automation with a better suit.
Read more about this at: SiliconANGLE