CrowdStrike has agreed to buy AI security startups Fiddler AI and Dreadnode. The Wall Street Journal reported that it is paying about $400 million for Fiddler and $95 million for Dreadnode, though CrowdStrike did not disclose prices. Fiddler tracks what AI agents do; Dreadnode tests AI models for flaws before they go live.

CrowdStrike has agreed to buy two artificial intelligence security companies, Fiddler AI and Dreadnode, as the cybersecurity giant tries to cover the AI agents and models that businesses are rapidly putting to work. The company announced both deals Tuesday at its Fal.Con Europe conference in Barcelona. CrowdStrike did not disclose prices, but The Wall Street Journal reported that it is paying around $400 million for Fiddler AI and about $95 million for Dreadnode.

What does Fiddler AI's observability software do?

The two startups do very different jobs. Fiddler makes observability software that tracks what AI agents do inside an enterprise and why they do it. Dreadnode automates security testing aimed at finding flaws in AI models, applications and agents before they go live. Fiddler was founded in 2018 and has raised about $99 million to date, according to PitchBook. Dreadnode, founded in 2023, has raised $14 million.

CrowdStrike plans to pay mostly in cash, with some stock subject to vesting conditions, according to the announcements from both companies. The deals are expected to close in the fourth quarter of CrowdStrike’s fiscal 2027, subject to customary closing conditions.

George Kurtz, CrowdStrike’s chief executive and founder, framed the purchase around accountability. “When a regulator or your board asks what an agent did, saying ‘the model did it’ is not an answer,” Kurtz said. “Fiddler is the flight recorder for every agent: the observability to explain exactly what an agent did and why.”

How the products fit into Falcon

CrowdStrike will sell Fiddler’s technology as Falcon Observe, part of its Falcon platform. The company says the tool will let customers evaluate agents before they are deployed, understand their actions and monitor what they do. That information will then feed into the broader platform as telemetry. Fiddler’s purpose-built models, according to the company, can cut evaluation costs by up to 98 percent.

How does Dreadnode test AI models for flaws?

Dreadnode brings a different kind of testing. It runs more than 100 known AI attack strategies, including jailbreaks and prompt injection, and targets models, agent reasoning loops, applications and MCP servers. Its findings map to frameworks such as the OWASP Top 10 for large language models, MITRE ATLAS and NIST’s AI Risk Management Framework. Dreadnode’s software will feed its testing results into Falcon Guardian as security policy recommendations, and its team will join CrowdStrike’s Cyber Superintelligence Lab.

Brad Palm, Dreadnode’s chief executive, pointed to a gap in traditional methods. “AI fails in ways traditional testing was never designed to catch, and the people best equipped to find those failures think like attackers,” Palm said.

A kill switch that is available now

What is CrowdStrike's AI Kill Switch?

Not everything announced this week is still in development. CrowdStrike also launched an AI Kill Switch within Falcon Guardian, its product for monitoring what agents do on devices. If an agent goes rogue, a defender can stop it where it runs and block it from launching again. The switch also cuts the agent’s access to the company network. An authorized administrator triggers it, and every action is logged. According to the release, the sensor stops agents at the gateway first and will stop them on the endpoint “soon.”

Rick Orloff, chief information security officer at Everpure, a CrowdStrike customer, offered an endorsement. “If one goes off course, the AI Kill Switch gives our security team the final say,” Orloff said.

The same day, CrowdStrike said it is bringing Falcon Guardian inside Anthropic’s Claude Cowork. The Falcon sensor runs within the virtual machine where Cowork’s agents execute code. The integration is in beta on macOS, and Windows support is expected later in November, the company said in a blog post.

The announcements came during a busy week for CrowdStrike in Barcelona. On Monday, the first day of the conference, the company announced security deals with Dell for its AI Factory with Nvidia, as well as with Dynatrace and CoreWeave. It also said 15 partners, including BT, Deutsche Telekom Security, Orange Cyberdefense and Telefónica, had joined Project QuiltWorks as it expands across Europe. That program uses frontier AI models to help customers find and fix security risks. The conference ends Wednesday.

Cet article a été rédigé avec l'assistance de l'IA.
News Factory APP - actualités agentiques pour booster votre SEO et AEO.