HiddenLayer Banks $100M to Lock Down AI Agents Before They Go Rogue
Austin-based AI security firm HiddenLayer has closed a $100 million Series B round, pushing its total funding past $155 million. The company, founded in 2022, builds security tooling specifically for AI systems rather than bolting traditional security approaches onto models that weren't designed with them in mind.
The platform covers the usual bases: AI supply chain security, attack simulation, discovery, and runtime protection. But the fresh capital is pointed squarely at agentic AI, the increasingly popular deployment model where autonomous agents take actions, write code, and make decisions with minimal human supervision.
That last part is where things get interesting and, frankly, a bit alarming. As enterprises hand more responsibility to AI coding agents that can write, review, and ship software on their own, the attack surface grows in ways that conventional security tools simply aren't built to handle. HiddenLayer's pitch is that you need AI-native security to monitor AI behaviour, catching manipulation, misuse, and unauthorised actions before they cause real damage.
The round was led by Delta-v Capital, with participation from Booz Allen Ventures, Microsoft's M12 venture fund, Morgan Stanley, and Ten Eleven Ventures. The Microsoft involvement is worth noting given how aggressively Redmond has been pushing Copilot and agentic tooling across its enterprise stack.
CEO Chris Sestito made the obligatory founder statement about being ahead of the curve, though in this case the underlying point is hard to argue with. Most organisations are still figuring out basic AI governance while simultaneously deploying agents that can take autonomous actions in production environments. That gap between adoption speed and security maturity is exactly what HiddenLayer is betting on.
Whether $155 million is enough to build the definitive platform for this problem, or whether the space fragments into a dozen niche tools, remains to be seen. But the money is clearly flowing toward anyone with a credible answer to the question nobody wants to ask out loud: what happens when your AI agent does something you didn't tell it to do?