HiddenLayer raises $100 million as enterprise AI security demand accelerates

HiddenLayer has raised $100 million in Series B funding to expand its security platform for AI models, agents and workflows, according to TechCrunch. Delta-v Capital led the round, with participation from Ten Eleven Ventures, Morgan Stanley, Microsoft’s M12, Booz Allen Hamilton and other investors.

The financing comes as companies move AI systems from experimentation into production, creating demand for controls that can detect prompt injection, agent manipulation, malicious tool use and compromised model files. HiddenLayer says it will use the capital to increase sales and distribution, continue engineering and research, and expand into Europe and the broader EMEA region.

HiddenLayer broadens its AI security platform

Austin-based HiddenLayer started by protecting machine-learning models from adversarial attacks, vulnerabilities and malicious code. Its product areas include discovery, runtime protection, attack simulation and supply-chain security.

Chief executive and co-founder Chris Sestito told TechCrunch that the company has expanded those capabilities rather than abandoned its original technology. The platform now addresses newer risks associated with generative AI and AI agents, including attempts to manipulate an agent or misuse the tools connected to it.

That change reflects a practical shift in enterprise deployments. A model is no longer operating in isolation when it can call software tools, retrieve information, make decisions or trigger business processes. Security teams therefore need visibility into the surrounding workflow, not only the model’s inputs and outputs.

Sestito described the company’s runtime security offering as comparable in concept to endpoint detection and response, but designed for AI systems. The comparison is useful for enterprise buyers evaluating where AI controls fit into existing security operations, although the source did not provide independent evidence of product performance or deployment scale.

Open-source models create a supply-chain problem

HiddenLayer is also targeting risks in open-source models and related AI files. Sestito said the company can parse and scan about 50 AI file frameworks to check whether a model is what it claims to be and to identify concealed models embedded within other files.

The concern is similar to software supply-chain security, but the artifacts and attack paths are specific to AI. A model downloaded from an external repository may be misrepresented, modified or packaged with hidden components. For product teams, that creates a need to verify model provenance before deployment and to monitor how models behave after they enter production.

The evidence available for those capabilities comes from HiddenLayer’s own description through TechCrunch. The company did not disclose independent testing, a public customer list or technical results that would allow buyers to compare its scanning approach with competing tools.

Funding follows reported revenue growth

HiddenLayer said its annual recurring revenue increased more than tenfold over the past year and is now in the tens of millions of dollars. Sestito told TechCrunch that more than 90% of that growth came from new customers signed during the period.

Those are vendor-reported business figures, not independently audited results. The company’s largest reported verticals are financial services and large technology companies building AI products. HiddenLayer also has contracts with the Department of Defense and the intelligence community, according to the report.

The company has not named a customer described as a leading frontier-model provider with more than 700 million weekly users. That description should not be treated as confirmation of any particular model developer or platform.

Market data cited by TechCrunch points to a broader spending increase. Gartner estimates that organizations will spend $2.83 billion on products for securing AI tools in 2026, up 83% from 2025, and forecasts spending of nearly $4.78 billion in 2027. These are analyst estimates, rather than evidence that every AI security vendor will capture equivalent growth.

Builders face a wider security perimeter

For AI builders, the financing highlights the growing importance of runtime controls alongside model evaluation and access management. An application that can invoke tools or act on behalf of a user needs policies governing which actions are allowed, how suspicious behavior is detected and when execution is stopped.

That is particularly relevant to teams deploying AI agents in finance, operations, customer support and internal knowledge workflows. A failure may not look like a conventional model error: an agent could follow a malicious instruction, call an inappropriate tool or pass harmful content into a downstream system. Security architecture must account for those chains of action.

Enterprise buyers will also need to determine whether specialist AI security products add capabilities not already available in cloud, identity or application-security platforms. Sestito acknowledged that some of HiddenLayer’s functions could eventually be bundled by companies such as Microsoft, OpenAI or AWS. He expects AI infrastructure providers to emphasize governance features such as discovery, identity and policy controls, while HiddenLayer focuses on adversarial behavior and runtime protection.

Competition is already forming around that boundary. TechCrunch noted that large cybersecurity companies including Cisco, Palo Alto Networks and Check Point often acquire technologies in adjacent categories, while Noma and Zenity have each raised more than $100 million. HiddenLayer’s challenge is to convert its current momentum into a durable position before those larger platforms close the product gap.

What to watch next

The first signal will be whether HiddenLayer can turn the new funding into measurable distribution outside the United States, particularly in Europe and EMEA. Enterprise security products often depend as much on channel relationships, compliance support and integration work as on detection technology.

Buyers should also watch for named customer references, independent assessments and clearer information about how HiddenLayer’s products integrate with cloud infrastructure, model registries and security operations centers. Those details would help distinguish broad AI security positioning from controls that work reliably in production.

Finally, the market will reveal whether runtime protection becomes a standalone budget category or is absorbed into broader enterprise AI governance and cybersecurity platforms. The answer will influence pricing, procurement ownership and the viability of specialist vendors.

Creati.ai perspective

HiddenLayer’s funding is a strong signal that AI security is moving from a research concern toward an enterprise deployment requirement. The most important part of the announcement is not simply the size of the round, but the company’s decision to extend model protection into the tools, files and workflows surrounding AI systems.

The harder test will be proof. Revenue growth and customer interest show commercial momentum, but enterprise buyers will ultimately judge HiddenLayer on false-positive rates, response speed, integration costs and its ability to prevent harmful actions without making AI applications unusable. The funding gives the company room to build that case; it does not establish it yet.