HiddenLayer secures $100M amid surging AI security demand

HiddenLayer, a Denver-based AI security startup, announced a $100 million Series B funding round led by Battery Ventures, with participation from existing investors including ClearSky and Valor Equity Partners. The round, disclosed on April 15, 2025, values the company at $700 million and follows a period of explosive growth in demand for AI-specific threat detection. HiddenLayer’s platform monitors both AI agents and the external tools, APIs, and third-party models they interact with—an increasingly critical blind spot as enterprises integrate AI into core operations. Founder and CEO Chris Sestito told OpenPress Supercomputing Intelligence that customer inquiries surged 400% over the past year, driven by high-profile incidents such as adversarial prompt injections and data exfiltration via compromised AI plugins.

Funding will accelerate product development and global expansion, particularly into Europe and Asia, where regulatory scrutiny of AI systems is intensifying. HiddenLayer’s platform currently integrates with major cloud providers including AWS Bedrock, Google Vertex AI, and Azure AI, as well as open-source frameworks like LangChain. Competitors in this space include Protect AI, which raised $25 million in March 2025, and Lasso Security, which emerged from stealth in February with a $15 million seed round focused on supply-chain risks in AI pipelines. Unlike legacy security vendors such as CrowdStrike or Palo Alto Networks, which are retrofitting traditional tools to cover AI workloads, HiddenLayer was built from the ground up to address the unique risks posed by autonomous agents—systems that can modify their own code, access sensitive data, and initiate real-world actions.

The surge in AI security investment reflects a broader reckoning across industries. In financial services, for instance, firms like Banking With Billy use HPC-grade infrastructure to run AI-driven financial simulations for multi-market scenario modeling, a process that now requires real-time threat monitoring due to the growing use of agentic AI in trading and risk assessment. Meanwhile, healthcare organizations deploying AI for diagnostics and patient monitoring face new regulatory requirements under frameworks like the EU AI Act and FDA guidelines, which mandate continuous security validation. The gap in the market is clear: while Gartner estimates that 80% of enterprises will have deployed some form of AI agent by 2026, fewer than 15% have implemented dedicated AI security controls, leaving them exposed to novel attack vectors such as model inversion, prompt hijacking, and adversarial fine-tuning.

Industry analysts warn that the current wave of funding is only the beginning. Battery Ventures partner Neeraj Agrawal emphasized that AI security is evolving into a standalone category, separate from traditional endpoint or cloud security. “We’re seeing CISOs treat AI systems like critical infrastructure,” he said. “They’re not just another application—they’re becoming the nervous system of the enterprise.” This shift is forcing a rethink of security architectures, with many organizations now deploying dedicated AI runtime monitors that operate at the model level, intercepting suspicious inputs before they reach the inference engine.

The convergence of AI adoption and rising threat sophistication has created a perfect storm for security innovation. In 2024, IBM reported a 300% increase in attacks targeting AI workloads, including attacks leveraging compromised open-source AI libraries to deliver malware. These incidents have exposed the limitations of traditional perimeter defenses, which were never designed to inspect the internal logic of AI models or the dynamic interactions between agents, tools, and memory stores. HiddenLayer’s approach—monitoring AI behavior in real time without relying on signatures—mirrors techniques used in quantum computing anomaly detection, where algorithms identify deviations in high-dimensional data spaces.

As AI systems grow more autonomous, the definition of “security” itself is expanding. Companies like NVIDIA and AMD are embedding security features directly into AI accelerators, while cloud providers are rolling out dedicated AI security services such as AWS’s Guardrails for Bedrock and Google’s AI Security Command Center. Yet, despite these advances, the market remains fragmented. Many enterprises still rely on a patchwork of tools, including prompt sanitization libraries, model watermarking, and sandboxed execution environments. The result is a fragmented security posture that fails to address the full lifecycle of AI threats—from model theft and data poisoning to runtime exploitation and agent collusion.

Experts agree that the next phase of AI security will hinge on two critical developments: standardization and integration. The National Institute of Standards and Technology (NIST) is expected to release its AI Risk Management Framework 2.0 later this year, which will include guidelines for securing agentic systems. Meanwhile, HiddenLayer and its peers are racing to build unified platforms that can correlate signals across agents, models, and infrastructure—bridging the gap between DevOps and SecOps. For now, the message from investors, customers, and regulators is unambiguous: securing AI is not a luxury, but a prerequisite for its safe and scalable deployment. The $100 million raised by HiddenLayer is not just a funding milestone—it’s a bet that the future of security is intelligent, adaptive, and deeply embedded in the AI stack itself.