HiddenLayer Raises $100 Million as AI Security Demand Surges

HiddenLayer has raised $100 million in new funding as enterprises accelerate spending on tools that secure AI models, agents and workflows from attacks, misuse and supply-chain threats. The Series B comes as the Austin-based startup says its annual recurring revenue has grown more than tenfold in the past year, underscoring how quickly AI security has moved from a niche concern to a major enterprise budget item.

The round arrives at a pivotal moment for the market: companies are deploying generative AI and agentic systems far faster than security teams can fully assess the risks, creating fresh demand for monitoring, runtime protection and guardrails that can stop malicious prompts, tool abuse and poisoned models before they cause damage.

Why HiddenLayer’s latest funding matters

HiddenLayer’s new financing is a signal that AI security is becoming a serious standalone category rather than an experimental add-on to existing cybersecurity products.

Three years ago, when the company raised its $50 million Series A, the industry was still debating whether there would be enough real-world attacks against AI systems to justify a large market. Today, that uncertainty has largely vanished. Enterprises are now deploying AI in production, regulators and customers are asking tougher questions about safeguards, and the security industry is racing to keep up with the pace of adoption.

Gartner expects spending on products that secure AI tools to reach $2.83 billion this year, up 83% from 2025, and nearly $4.78 billion next year. That kind of growth helps explain why investors are still willing to write large checks even as the market remains crowded and fast-evolving.

What HiddenLayer does

HiddenLayer builds security software designed to protect AI models, agents and AI-driven workflows from adversarial attacks, vulnerabilities and malicious code injection.

The company’s core offerings include discovery, runtime protection, attack simulation and supply-chain security. Over the past year, it has expanded those products to address newer threats tied to prompt injection, agent manipulation and the misuse of external tools connected to AI systems.

How the company describes the shift

HiddenLayer chief executive Chris Sestito says the company did not need to reinvent itself to follow the AI boom. Instead, it had to broaden the same security principles it already used for machine-learning systems and apply them to generative AI and agentic workflows.

Sestito said that inference remains inference, whether it is powering a traditional machine-learning model, a generative AI tool or an agentic workflow, and that the company’s technology still applies across those environments. In his view, HiddenLayer has not pivoted so much as expanded its coverage from older model security to a wider AI stack.

That distinction matters. Many startups have rushed into AI security with products narrowly focused on chatbots or prompt filters. HiddenLayer is positioning itself as infrastructure for the broader enterprise environment, where AI systems are increasingly embedded in daily operations and connected to internal data, APIs and third-party tools.

How big is the AI security opportunity?

The market is growing quickly because companies are worried about both obvious and subtle failures in production AI systems.

On the obvious side, enterprises need to stop attackers from hijacking models, stealing data or injecting malicious instructions. On the subtler side, they need to prevent AI agents from taking unsafe actions, using the wrong tools or following hidden commands buried in data, documents or web content.

HiddenLayer argues that runtime security is becoming especially important as AI moves from demos and pilots into business-critical deployments. The company compares the need for AI runtime defense to endpoint detection and response, or EDR, in traditional cybersecurity: a control layer that watches systems continuously and intervenes when behavior becomes suspicious.

Who is using HiddenLayer?

HiddenLayer says its biggest customer groups are financial services firms and large technology companies building AI products.

The startup also works with the Department of Defense and elements of the U.S. intelligence community, indicating that public-sector interest in AI defense is growing alongside commercial demand.

Sestito also said one customer is a frontier model provider with more than 700 million weekly users. While he did not name the company, the description strongly suggests a major AI platform operator such as OpenAI or Anthropic.

Why regulated industries are moving first

Financial institutions and defense-related organizations tend to adopt security controls earlier because their risk tolerance is lower and their compliance obligations are higher. AI systems that generate advice, analyze sensitive data or interact with internal systems can create legal, financial and operational exposure if they behave unpredictably.

That makes AI security a natural early spend area for organizations already accustomed to spending heavily on threat detection, governance and monitoring.

What’s in the new products?

HiddenLayer is leaning into a broader set of attack surfaces that have emerged as AI systems became more complex.

The company says it now scans about 50 AI file formats and frameworks to help verify that open-source and open-weight models are what they claim to be. That matters because malicious actors can hide unwanted behavior inside model files, embed deceptive components or distribute altered versions of otherwise legitimate tools.

The startup is particularly focused on risks that come with open-source AI, where developers may download and run models from outside vendors without full visibility into their contents or provenance.

Sestito said HiddenLayer is watching for cases where a model appears to be one thing but contains something else underneath, including hidden models nested inside other models.

That sort of threat is not theoretical. As open-weight ecosystems expand, security teams are increasingly responsible for verifying that model artifacts, dependencies and attached tools have not been tampered with before they are deployed.

Who backed the Series B?

The funding round was led by Delta-v Capital and included participation from Ten Eleven Ventures, Morgan Stanley, Microsoft’s M12, Booz Allen Hamilton and other investors.

The investor mix reflects a common pattern in AI security: specialist cyber funds are joined by strategic backers that have a direct interest in protecting enterprise and government AI deployments.

Why strategic investors matter

Backers like Microsoft and Booz Allen Hamilton can offer more than capital. They provide market access, customer relationships, credibility in regulated environments and a better view into how AI security demand is changing across large enterprises and public agencies.

For HiddenLayer, that matters because many AI security buyers are still in the early stages of defining what they need, how much they should spend and where security should sit in the AI stack.

How will HiddenLayer use the money?

The company says the capital will go toward sales and distribution as well as continued investment in engineering and research.

It also plans to expand into Europe and the broader EMEA region, suggesting that HiddenLayer sees the next phase of growth as international rather than purely domestic.

That expansion will likely require more local partnerships, compliance work and product adaptation, especially as European enterprises ask tougher questions about model governance, data handling and AI accountability.

What risks does HiddenLayer face?

HiddenLayer is entering a market that is attracting both startups and established cybersecurity giants.

Large security vendors such as Cisco, Palo Alto Networks and Check Point have a long history of acquiring promising security technologies when they become strategically important. That means HiddenLayer could eventually face pressure from both sides: buying interest from incumbents and competition from startups trying to carve out adjacent niches.

Other venture-backed players, including Noma and Zenity, have already raised more than $100 million each to tackle overlapping parts of the same market.

• Big security firms may choose to build or buy AI defense features.
• Cloud and AI platform providers may bundle baseline protection into their stacks.
• Startups are competing to define categories before the market consolidates.
• Enterprise buyers want simpler, more integrated controls rather than a patchwork of tools.

Will AI platforms absorb this market?

Some of HiddenLayer’s functionality could eventually be folded into the products of larger platform providers, and Sestito acknowledges that possibility.

He expects AI infrastructure to become richer over time, with governance features such as identity, policy controls and discovery becoming standard parts of the stack. In that scenario, some security capabilities may be built directly into platforms from Microsoft, OpenAI, AWS and others.

Even so, HiddenLayer believes there will still be room for a specialized vendor focused on deeper protection, runtime controls and adversarial testing. The company’s thesis is that as AI becomes more embedded in enterprise systems, security requirements will become more specialized, not less.

Why differentiation will be hard

That argument has merit, but it comes with a challenge: platform giants can move fast when a feature becomes table stakes. HiddenLayer will have to prove that enterprises want a dedicated AI security layer rather than enough built-in tools to satisfy most buyers.

Its advantage today appears to be timing, technical specialization and a growing list of paying customers. The question is whether those advantages will endure once the market matures and vendors converge on similar feature sets.

A short timeline of HiddenLayer’s rise

HiddenLayer’s growth mirrors the broader evolution of AI security from a speculative idea to an enterprise budget line.

What this says about the AI market overall

HiddenLayer’s fundraise is part of a larger pattern across enterprise AI: adoption is moving quickly, but trust remains incomplete.

Businesses want the productivity gains from AI agents, copilots and automated workflows, yet they are increasingly aware that these systems can be manipulated in ways traditional software cannot. That tension is creating a market for security tools built specifically around AI behavior, data provenance and runtime oversight.

For investors, the logic is straightforward. Whenever a new computing layer becomes essential, a security layer usually follows. HiddenLayer is betting that AI will be no different.

The company now has the balance sheet to prove that thesis, but the next phase will be harder than the first. It must convert momentum into durable market share, defend its niche against larger rivals and show that AI security is not just a hot category, but a lasting one.

What comes next for HiddenLayer?

HiddenLayer’s immediate priorities are growth, product depth and international expansion.

Longer term, the startup is aiming to “scale vertically alongside artificial intelligence,” in Sestito’s words, and later expand into adjacent parts of cybersecurity that increasingly rely on AI themselves. That is a broad ambition, but it reflects the direction of the market: the more AI runs core business processes, the more security vendors will need to understand AI as both the threat and the defense.

For now, HiddenLayer is one of the clearer examples of how quickly the AI security sector has shifted from a future concern to a present-day business opportunity.