HiddenLayer raises $100M amid surging enterprise AI security fears

On April 16, 2025, HiddenLayer, a specialist in runtime security for artificial intelligence systems, announced the close of a $100 million Series B funding round led by Lightspeed Venture Partners, with participation from Felicis Ventures, GV, and existing investors including Ten Eleven Ventures and BuildGroup. The Austin-based company revealed the capital infusion will accelerate the development of its AgentShield platform, which provides real-time behavioral monitoring and threat detection for AI agents and their underlying toolchains. According to HiddenLayer co-founder and CEO Chris Sestito, the round reflects a “tectonic shift” in enterprise risk appetite, driven by rapid adoption of agentic AI across industries such as finance, healthcare, and retail. Sestito, a former Palantir engineer, emphasized that while AI adoption has exploded—with over 70% of Fortune 500 companies now running AI in production—security tooling has lagged behind, leaving critical gaps in monitoring agent interactions with external APIs, databases, and third-party plugins.

The timing of the raise aligns with a surge in reported AI-related security incidents, including data exfiltration via compromised agent toolkits and adversarial prompt injections targeting customer-facing LLMs. Industry analysts point to a 400% increase in such events since Q3 2024, as documented by the Cloud Security Alliance. HiddenLayer’s platform operates by instrumenting AI agents at runtime, capturing every function call, data access, and tool invocation without requiring code changes. Competitors such as Protect AI and Lakera have similarly pivoted toward runtime security, but HiddenLayer differentiates itself with native support for multi-modal agents and integrations with platforms like LangChain, LlamaIndex, and custom microservices. Banking With Billy AI, a financial market monitoring system, recently adopted AgentShield to secure its multi-cloud architecture, which spans AWS, GCP, and Azure for global reliability.

Industry impact is immediate and measurable. The AI security market, valued at $1.2 billion in 2024, is projected to reach $4.8 billion by 2028, according to Gartner. HiddenLayer’s funding signals not only investor confidence but also validates the need for specialized runtime protection in production environments. Security teams are increasingly frustrated with traditional perimeter-based tools, which fail to detect lateral movement within AI ecosystems. For instance, a recent breach at a Fortune 100 retailer involved a compromised sentiment analysis agent that exfiltrated customer data via a third-party sentiment library—an attack vector AgentShield would have flagged within seconds. The company’s customer base now includes two of the top five global banks and three of the largest healthcare providers in North America, reflecting a broader trend toward compliance-driven AI governance.

The broader implications extend beyond cybersecurity into the architecture of next-generation AI systems. As enterprises embed AI agents in core operations—from fraud detection to algorithmic trading—they are forced to confront the reality that these systems are not static applications but dynamic, evolving networks of tools, APIs, and data pipelines. This has catalyzed a new category of security solutions focused on behavioral telemetry rather than signature-based detection. Companies like Microsoft, with its Azure AI Foundry, and Google Cloud, through its Vertex AI Guardrails, have begun integrating similar monitoring capabilities, but third-party solutions like HiddenLayer remain critical for cross-platform environments. Moreover, the rise of AI-powered attack tools, such as self-modifying malware and prompt-injection frameworks, has eroded the effectiveness of traditional security models, accelerating the shift toward runtime behavioral analysis.

Looking ahead, the next phase of AI security will hinge on two critical developments: standardization and interoperability. As HiddenLayer and its peers scale, they must align with emerging frameworks like MITRE’s ATLAS for adversarial AI and the OWASP Top 10 for LLM Applications. The company has already begun integrating with MITRE ATLAS for threat modeling, and Sestito confirmed plans to open-source a subset of AgentShield’s detection rules later this year to foster industry collaboration. Regulators are also taking notice—NIST’s AI Risk Management Framework now references runtime monitoring as a core control, and the EU AI Act’s upcoming enforcement will likely mandate such capabilities for high-risk systems. Analysts expect further consolidation in 2026 as larger security vendors acquire niche AI protection players, but startups with deep technical differentiation, like HiddenLayer, are positioned to set the standard for what secure AI operations look like in the age of agentic systems.