HiddenLayer Raises $100M for Runtime Controls on Enterprise AI Agents

HiddenLayer has raised a $100 million Series B to expand security software for enterprise AI systems, with a particular focus on watching AI agents while they operate. Delta-v Capital led the round, joined by Ten Eleven Ventures, Morgan Stanley, M12, Microsoft's venture fund, and Booz Allen Ventures.

The company plans to spend the capital on product development, engineering, research, sales, distribution, and expansion into Europe and the wider EMEA region. Its immediate product priorities are Agentic Runtime Security and Agent Harness Security, which it positions for AI coding agents.

Security shifts from the model to its actions

HiddenLayer already sells tools for discovering AI assets, protecting systems at runtime, simulating attacks, and securing the AI supply chain. It has broadened that scope to cover prompt injection, agent manipulation, and malicious tool use as companies put generative AI and agents into production.

The distinction is timing. HiddenLayer says its runtime tools give organizations visibility into agent behavior and can flag or stop manipulation, tool misuse, and unauthorized actions as they happen. Agent Harness Security extends that approach to agents that write, review, and ship code.

A larger budget, and a harder product boundary

HiddenLayer also targets model files, especially open-source and open-weight models that organizations bring into their systems. Chief executive Chris Sestito said the company scans about 50 AI file frameworks for tampered artifacts, including files that may conceal another model or differ from their labels.

The company reported that annual recurring revenue grew more than tenfold over the past year to the tens of millions of dollars, though it did not disclose an exact figure. It also said it signed more than 50 new platform customers across sectors including banking, insurance, pharmaceuticals, technology, and government.

The question behind the financing

The new funding also supports sales and distribution as the company tries to establish a standalone role. Sestito acknowledged that infrastructure providers such as Microsoft, OpenAI, and AWS could eventually bundle parts of AI security into their platforms. He expects those platforms to focus more on governance features such as discovery, identity, and policy controls, while HiddenLayer supplies specialized protections.

That leaves HiddenLayer with a commercial test as well as a technical one: whether enterprises treat runtime protection as a separate control layer, rather than a capability folded into their existing AI and cybersecurity platforms.