CrowdStrike Plans to Buy Dreadnode to Help Businesses Test AI Agents Before They Go Live

CrowdStrike announced at Fal.Con EU 2026 that it intends to acquire Dreadnode. The deal would bring automated AI security testing to the Falcon platform, so customers can check the security of AI systems before they go into production. CrowdStrike says Dreadnode will be generally available in Q4 FY27.

The company argues that many businesses deploy AI models and agents before confirming they are secure. If an agent ships with a weakness that allows jailbreaking or manipulation, an attacker may be able to reach the sensitive systems and data that the agent can access. CrowdStrike says testing built for conventional software does not find these problems, because AI agents and large language models behave in a dynamic, probabilistic way rather than consistently. It names prompt injection, jailbreaking and agent tool manipulation as risks that can lead to customer data loss, financial loss and brand damage.

According to CrowdStrike, users describe in plain language what they want to test. Dreadnode then runs more than 50 attack techniques, including jailbreaks, prompt injection, and multi-turn and agent-specific attacks. These are run against AI models, model repositories, applications, agents and MCP servers before they reach production, so fixes can be applied beforehand. This summary is based on the opening of CrowdStrike's announcement, not the full article.