CrowdStrike to buy Fiddler AI & Dreadnode in AI push

CrowdStrike plans to acquire Fiddler AI and Dreadnode as part of a broader push into AI security and oversight.

It also introduced an AI Kill Switch designed to let defenders terminate rogue agents, and announced partnerships with Anthropic, EY, AWS and Lovable.

Fiddler AI focuses on model monitoring and observability, while Dreadnode works on automated AI security testing. CrowdStrike said the additions are intended to strengthen how organisations track, assess and govern AI systems and agents.

The announcements show how cybersecurity groups are widening their scope as businesses deploy AI tools more deeply across operations. Security vendors are increasingly trying to address not only traditional cyber threats, but also risks linked to autonomous software agents, model behaviour and misuse of AI systems.

Acquisition plans

With its planned purchase of Fiddler AI, CrowdStrike is moving into agent observability, an area focused on how AI systems behave and whether their outputs remain within expected limits. The planned acquisition of Dreadnode adds technology aimed at testing AI systems for weaknesses and trust issues before they are widely deployed.

Neither financial terms nor expected closing timelines were included in the announcement. CrowdStrike framed the transactions as part of a wider strategy around what it described as AI trust.

Kill switch

The new AI Kill Switch centres on direct intervention when an AI agent behaves in a way defenders judge to be unsafe or unauthorised. In practice, that emphasises stopping an automated process after it has begun operating, rather than only screening it before use.

That reflects a growing concern among companies adopting agent-based AI tools: once given access to data, applications or workflows, those systems may act in unintended ways or exceed approved boundaries. Vendors across the sector are developing controls intended to monitor, test and, where necessary, halt those actions.

Partner ties

The partnership announcements extend the same theme into the broader technology and services market. CrowdStrike said it is extending protection for Anthropic's Claude for Enterprise use, building new AI security work on AWS, and working with Lovable on security for AI-built applications.

EY was also named as part of the expansion. CrowdStrike said EY is building its frontier AI and cyber resilience business on the Falcon platform and has committed to training its cyber workforce on it.

The combination of acquisitions, product changes and ecosystem deals suggests CrowdStrike is trying to position itself across several layers of the AI security market, including model oversight, security testing, runtime intervention, and partnerships with cloud, consulting and application-focused groups.

Competition in that market is intensifying as cybersecurity companies and newer AI-focused specialists seek to define standards for how organisations manage risk in AI deployments. Questions around observability, testing and operational control have become more pressing as companies move from limited pilots to broader use of AI agents in internal and customer-facing work.

CrowdStrike's latest moves also indicate that AI security is becoming a commercial category in its own right, rather than a feature folded into general cybersecurity products. By pursuing acquisitions and alliances at the same time, the company is seeking both additional technology and broader market reach as customers weigh how to govern AI systems already entering daily business processes.

The package of announcements is aimed at giving organisations more ways to secure, monitor and control AI use across enterprise environments.