Share with your CISO
HiddenLayer has closed a $100 million Series B led by Delta-v Capital, a firm built specifically around the gap between traditional cybersecurity and AI-native threats. The capital goes toward HiddenLayer’s Agentic Runtime Security platform and a new product called Agent Harness Security, which targets autonomous AI coding agents operating in production. DLA Piper advised Delta-v. The round signals that investor conviction around securing AI agents at runtime, not just at deployment, has crossed from niche thesis into funded reality.
What this means for your business
Most enterprise security stacks were designed before AI agents existed as a production concern. The organizations most exposed here aren’t necessarily the least mature, they’re the ones that moved fastest on agentic AI and now have autonomous coding agents running in environments their security tooling wasn’t built to monitor. HiddenLayer’s runtime focus, watching what an agent does while it runs rather than scanning it before launch, addresses a class of threat that static analysis and traditional endpoint tools simply don’t see.
A $100 million Series B doesn’t validate a product. It validates a market. Delta-v’s entire thesis is the wedge between legacy cyber and AI-specific risk, and backing HiddenLayer at this size means they believe enterprises are ready to buy a dedicated layer for AI agent security rather than waiting for incumbents like CrowdStrike or Palo Alto to retrofit coverage. That’s a real bet, and it’s worth watching whether the incumbents respond with acquisitions or product announcements in the next 12 months, because that response is the actual signal about whether this category stays independent.
The budget question this reshapes isn’t whether to fund AI security, it’s whether your existing security vendors’ roadmaps will cover agentic runtime risk on a timeline that matches your AI deployment schedule. If your organization is already running autonomous coding agents in production, the renewal conversation with your current security stack deserves a direct question about runtime agent visibility, and the answer will tell you whether you have a gap or just a gap in the marketing materials.
Concept deep-dive: Agentic Runtime Security
Runtime security means monitoring and enforcing policy on a system while it’s actively executing, not before or after. For AI agents, which can chain decisions, call external APIs, and write or execute code autonomously, the threat surface only exists at runtime. Traditional security scans the model or the code before it runs. Agentic runtime security watches behavior as it happens, the AI equivalent of an endpoint detection agent sitting inside a running process rather than checking files at the door.
Based on reporting from DLA Piper advises Delta-v Capital in Series B investment advancing enterprise AI security, originally published 2026-09-07 02:35:00.
