CrowdStrike announced at its Fal.Con Europe conference in Barcelona that it is acquiring two AI security companies: Fiddler AI and Dreadnode. The company did not disclose prices, but according to reports in the financial press the deals total about $495 million: about $400 million for Fiddler and about $95 million for Dreadnode. The deals are expected to close in CrowdStrike's fiscal fourth quarter of 2027.
The two companies complement each other. Fiddler builds tools for monitoring AI systems, and its technology will be integrated into CrowdStrike's platform as Falcon Observe: tracing AI agents' decisions, monitoring what they actually do and logging for compliance. Dreadnode covers the other side: it tests AI systems before they go live against more than 100 attack techniques, such as jailbreaks, prompt injection and attacks aimed at agents. Along with the acquisitions, CrowdStrike launched a 'kill switch' that can stop an AI agent in real time.
The deals come in a week when the field got a painful reminder of why it is needed: WorkOS reported that an autonomous attack agent reached its customer data. CrowdStrike's logic is simple: organizations are deploying AI agents with access to systems and data, and nobody knows exactly what they are doing. Whoever sells the tool that answers that question will sell it to every large organization.
The Israeli angle is interesting precisely because of its absence. AI security is one of the hottest areas in Israel's cyber industry, and earlier this month Calcalist reported that CrowdStrike itself is in talks to acquire Israeli startup Orion Security for about $500 million. But it closed these two large deals with American companies. The field is no longer reserved for Israelis, and buyers choose the technology, not the address.
There is also an open question about the prices. $495 million is a lot of money for two young companies, in a market where it is still unclear how many customers are willing to pay. Cyber giants today are buying the fear of AI agents, not proven revenue. If enterprise adoption of agents slows, some of these deals may look like peak prices in two years.
