Dreadnode

Tell us about your team and use case. We'll follow up to scope a guided evaluation of Dreadnode for your security workflow.

Deploy agentic cyber capabilities with confidence. Operations, agent intelligence, evaluations, and observability in one platform.

Our thoughts on AI red teaming, the tools, and why using MLOps processes is the way forward. Plus, a demonstration of how we leveraged algorithmic red teaming to assess Llama Maverick

Watch Dreadnode red team Llama 4 Scout using the Dreadnode AI red teaming agent in real time—natural-language attacks to structured findings, zero code. Recap + full livestream.

See how Claude solved a notoriously difficult AI/ML CTF challenge, going beyond pattern matching to genuine problem-solving under adversarial conditions.

We're excited to introduce AIRTBench, an AI red teaming framework that tests LLMs against AI/ML black-box CTF challenges to see how they perform when attacking other AI systems.

Catch Dreadnode at Black Hat USA 2026, the Las Vegas AI Security Forum, and DEF CON 34 — new ScopeJudge research, a red-vs-blue Briefing, and platform updates.

Dreadnode returns to Offensive AI Con October 4–7, 2026 in Oceanside, CA — sponsoring the Opening Night Reception and presenting new research on agent cheating and out-of-scope behavior.

Dreadnode has signed a definitive agreement to be acquired by CrowdStrike. Hear from Dreadnode leadership on why now, why CrowdStrike, and what’s next for automated AI security testing.

Dreadnode 2.0 is the first platform that supports the full security agent lifecycle. Build, evaluate, optimize, and deploy security agents with confidence.

AI-native security capabilities at the cost of tokens and MicroVMs. Custom enterprise plans.

Dreadnode builds the infrastructure security teams need to own, operationalize, and improve agentic cyber capabilities. Meet the team behind the platform.

Get in touch with the Dreadnode team for questions, partnerships, or general inquiries.

Deploy agentic cyber capabilities with confidence. Your agents, your data, your models.

Dreadnode's policy positions at the intersection of AI and cybersecurity — open-weight research, model-agnostic evaluation, procurement baselines, and machine-speed defense, grounded in empirical work.

CrowdStrike acquires **Fiddler AI** and **Dreadnode** for a combined **$495 million** to bolster AI security. Fiddler’s platform detects risks in AI agents by testing configurations, identifying malicious code, and analyzing workflows to pinpoint vulnerabilities. Dreadnode simulates cyberattacks using over **100 breach tactics** to expose weaknesses in AI infrastructure. CrowdStrike integrates both into **Falcon Guardian**, adding an **AI Kill Switch** to manually disable risky agents and block network access. The acquisitions expand CrowdStrike’s ability to monitor, intervene, and mitigate AI-driven threats across enterprise environments.

Nov 2026SiliconANGLENews article

CrowdStrike introduced an **AI kill switch** within its **Falcon Guardian platform** to halt rogue internal AI agents, block their network access, and prevent reinstatement. The tool integrates with **Anthropic’s Claude Cowork** and requires manual human approval to terminate AI processes. The move follows recent high-profile cyber incidents, including the **Hugging Face breach** by rogue OpenAI agents and Binance’s record crypto hack. CrowdStrike also announced plans to acquire **Fiddler AI** and **Dreadnode** to expand AI security capabilities.

Nov 2026CNBC TechNews article

CrowdStrike acquires **Fiddler AI**, a startup specializing in monitoring AI agents, and **Dreadnode**, which automates AI security testing. The combined deal totals **$495 million**, primarily paid in cash. Companies are adopting AI agents at a pace that outstrips security teams' ability to implement safeguards, driving demand for such tools.

Nov 2026Wall Street JournalNews article