Startups & Funding
CrowdStrike to acquire AI red-teaming startup Dreadnode
CrowdStrike has signed a deal to acquire Dreadnode, a startup that automatically tests AI models and agents for security flaws, folding its technology into the Falcon platform.
CrowdStrike announced the agreement Monday at Fal.Con Europe 2026 in Barcelona, framing the deal as a fix for a gap in how companies secure AI: traditional security testing assumes software acts identically every run, whereas AI models and agents behave probabilistically, can be tricked into jailbreaks or steered off course by injected prompts, and break in ways that older testing approach was never built to notice. “Enterprises are deploying AI faster than they can prove it is secure,” said CEO and founder George Kurtz, adding that a jailbroken model or a manipulated agent usually goes undetected because most organizations have no way to surface those flaws before production.
Once the deal closes, a user will be able to tell Dreadnode what to test in everyday language, after which it runs over 100 established attack techniques for AI systems — jailbreak attempts, injected-prompt exploits, and attacks spanning multiple turns or targeting agents specifically — against the underlying models, an agent’s reasoning loop, the applications built on them, and the MCP servers that connect everything. Findings come back ranked by severity and mapped to standards such as the OWASP Top 10 for LLMs, OWASP Agentic Security, MITRE ATLAS, and NIST AI RMF, with exportable reports teams can use to decide on deployment. CrowdStrike expects the capability generally available by its fourth fiscal quarter of FY27.
Dreadnode’s team will join CrowdStrike’s Cyber Superintelligence Lab, and its technology will also feed into SafeMind, CrowdStrike’s system for measuring the security and cyber capabilities of frontier AI models. “We founded Dreadnode to expose those weaknesses before AI reaches production,” said co-founder and CEO Brad Palm. The acquisition is expected to close in CrowdStrike’s fourth fiscal quarter of FY27, subject to customary closing conditions.
Why it matters
The deal signals that automated adversarial testing of AI systems is becoming a standard line item for enterprise security vendors, not a niche research exercise, as companies race to deploy agents before they can fully verify them.