When AI security startup HiddenLayer raised its $50 million Series A three years ago, the question was whether AI threats would materialise enough for a real market. Now, security companies are scrambling to build tools to monitor AI agents and their tools, as the risk of agents going rogue is real. HiddenLayer, which makes tools to protect AI models and workflows, has seen its annual recurring revenue grow more than 10x over the past year, with financial services and large tech companies being its largest verticals. The startup has now raised $100 million in a Series B funding round, which it plans to use to expand its engineering and research, as well as to push into Europe and EMEA.

The biggest change for HiddenLayer has been to extend its products to address prompt injection, agent manipulation, and malicious tool use, as AI deployments grow common across businesses. The company is particularly focused on runtime security, which it sees as a priority as AI deployments become more widespread, much like traditional endpoint detection and response solutions for AI. HiddenLayer is also tackling the new challenge of exploiting open-source models, where it scans about 50 different AI file frameworks to ensure that the model being used is the one it claims to be.

While some parts of the AI security products HiddenLayer offers could eventually be bundled into the platforms built by companies like Microsoft, OpenAI, and AWS, the startup expects AI infrastructure to grow towards governance features such as discovery, identity, and policy controls, rather than the kind of tools it builds. For now, HiddenLayer’s job is to scale vertically alongside artificial intelligence, and eventually expand horizontally into parts of cybersecurity that increasingly depend on AI. However, the startup first needs to prove it can turn its head start into an enduring business before the rest of the industry catches up.