Dreadnode, an early-stage startup specializing in offensive AI security, recently secured $14 million in Series A funding. The investment round was backed by a diverse group of investors, including Decibel, Next Frontier Capital, In-Q-Tel, Sands Capital, and Indie VC. This funding comes amid rising concerns over the security of artificial intelligence systems, as AI technology becomes increasingly integrated into critical infrastructure and various industries. The company’s focus is on developing new products aimed at stress-testing AI models for potential vulnerabilities and ensuring their security.
The company was founded by Will Pearce, former NVIDIA AI red-team lead, and Nick Landers, ex-NetSPI VP of Research.
They aim to address the lack of tools designed to secure AI-driven systems by creating solutions that allow businesses to simulate potential AI vulnerabilities in a controlled environment. Their products, Strikes and Spyglass, form the core of their platform. Strikes is designed as a training ground for AI agents, offering real-world attack scenarios to help developers evaluate and improve the resilience of their models. Spyglass, on the other hand, focuses on testing deployed AI systems, auditing live models for weaknesses such as prompt injections, data poisoning, or other vulnerabilities that might compromise their functionality.
Dreadnode’s approach is rooted in offensive machine learning, providing AI developers and cybersecurity professionals with the tools to simulate and defend against AI-specific cyber threats.
Strikes allows practitioners to simulate various attack methods against AI models during training, while Spyglass enables continuous assessment of deployed AI applications. By incorporating the latest advancements in adversarial machine learning, Dreadnode aims to help organizations identify weaknesses early in the development process and while the models are in use, ensuring that their AI applications are resilient to emerging threats.
In addition to these tools, Dreadnode offers Crucible, a sandbox environment where security professionals can hone their skills in AI red-teaming and offensive AI tactics. The company’s founders believe that AI’s potential in cybersecurity should be harnessed not just for defensive purposes but also in offensive operations to identify and exploit vulnerabilities before adversaries can do so. With the new funding, Dreadnode plans to accelerate the development of its products and expand its market presence, positioning itself as a key player in the rapidly evolving field of AI security.
