Amazing work by our CrowdStrike team working together with the Federal Bureau of Investigation (FBI) to disrupt laptop farms operating across the U.S. maintained by DPRK-nexus adversary FAMOUS CHOLLIMA to fraudulently employ operatives at American companies while concealing their true location and identity.
These arrests are significant, these laptop farms enabled FAMOUS CHOLLIMA operatives to pose as legitimate remote workers and siphon illicit wages back to their regime while maintaining persistent access to victim organizations.
Great read here.
https://lnkd.in/eNGWrtAR
Wow, thank you to everyone that attended Fal.con 2026 this morning, that was a lot of fun!!!
Software has become an actor.
AI is the biggest technology transformation ever, and there is opportunity for adversaries AND defenders. The question now is what happens when millions of AI agents act together at machine speed.
AI changes who or what can act; fundamentally AI changes how we trust. AI changes how we defend.
Operationalizing AI security comes down to three things: visibility into AI agents, stopping supply chain attacks at the endpoint, and identity-based access decisions for agents.
It was a privilege to be able to announce a heap of innovations and show them off on stage!!
Falcon Guardian:
Runtime visibility across the AI estate. It secures agents on the endpoint at runtime, with activity detected by the Falcon sensor in real time.
AI Gateway:
Lets teams decide what models they can use and apply those controls wherever AI runs, in the cloud and on the endpoint.
Real-Time Supply Chain Attack Protection:
Software supply chain risk converges on the endpoint. We used to defend at the perimeter; now agents are in your environment. It stops malicious packages before they can execute. It’s not another console or another build. It’s part of the Falcon platform.
Agentic Identity Provider:
AI agents create a new identity problem. Trust cannot stop at authentication, and agent access cannot be a one-time decision. Within Falcon Next-Gen Identity Security, it provides trusted identities for agents and gives organizations the ability to provide access on demand.
Agentic SOC:
Autonomous attacks killed breakout time, and agents exploit vulnerabilities in real-time. Investigations at machine speed are a requirement now. But speed only matters if you can trust the outcome. You connect the systems and the agents your team needs.
SafeMind:
Across the AI estate, defenders need Frontier AI. Defenders now have it.
CrowdStrike has created a speed advantage for the defender community. #CrowdStrike #Fal.con2026

Thank you Barcelona!!! What a crowd at Fal.Con Europe.
First, a massive welcome to the Fiddler.ai and Dreadnode teams. Welcome to the CrowdStrike family!
I always enjoy spending time with customers in this region and appreciate the quality of our conversations. And right now the conversation is dominated by AI.
An agent doesn’t have to be malicious to be dangerous. It can follow a legitimate instruction, use legitimate credentials, access legitimate systems, and still do something nobody intended. We’ve spent decades securing what software can do. Now we have to secure what software decides to do.
It was an honor to show the innovation we’re delivering to solve these challenges on stage!
Falcon Guardian:
So many of our customers are already running Guardian, and we keep delivering what our customers need to run AI securely.
We secure agents where they run, on the endpoint, at runtime. We see the prompts, enforce policy, and block dangerous actions. We even put the Falcon sensor inside Anthropic’s Claude Cowork environment so customers can run it securely.
AI Kill Switch:
So many customers talking about this at Fal.Con Europe! You cannot ask an agent to stop itself. The control has to live outside the agent. One action that stops an agent cold, everywhere it is running, all at once.
Falcon Observe:
That’s where Fiddler comes in. Guardian secures what agents do. Falcon Observe tells you if they’re doing the right thing, before deployment and in production. Every agent. Every decision. Every action.
Agentic Identity Provider:
An agent needs its own identity, not unlimited access borrowed from ours. Every agent gets its own identity. When the risk changes, access changes. Automatically.
Automated AI Security Testing:
With Dreadnode, you test your AI the way an adversary would. What you learn before it goes live protects it once it’s running. This is so important.
Agentic SOC:
Agents are the new insider. They have real access and they don’t break a rule, so there’s nothing for your tools to catch. Your agents need to investigate together. That’s the agentic SOC.
Our job in cybersecurity is not to slow down the future. Our job is to make the future possible.
#CrowdStrike ##FalConEurope

Worth reading this blog from OpenAI! This is a milestone moment for AI security. The question is no longer whether AI can execute sophisticated cyber operations, it's whether governance, containment, and detection capabilities are evolving just as quickly, and how you as a defender are keeping pace. This is one to follow thats for sure!
https://lnkd.in/gmyH-g4a