HiddenLayer secures $100M to shield enterprise AI from threats

HiddenLayer, the Austin-based startup focused on securing enterprise AI deployments, announced the close of a $100 million Series B on June 10, 2024. Led by Delta-v Capital with participation from Ten Eleven Ventures, Morgan Stanley, Microsoft’s M12, Booz Allen Hamilton, and additional strategic investors, the round values the company at more than $500 million. HiddenLayer’s platform monitors real-time inference pipelines, detecting adversarial prompts, model theft, and data poisoning attacks without requiring access to proprietary model weights. According to CEO Chris Sestito, the company now protects over 100 enterprise AI environments across finance, healthcare, and defense, with customers including a top-three U.S. bank and a Fortune 50 manufacturer. The funding will accelerate R&D in runtime AI threat detection and expand global go-to-market operations, particularly in Europe and Asia-Pacific where regulatory scrutiny of AI safety is intensifying.

The round comes just six months after HiddenLayer’s $15 million Series A in December 2023, reflecting rapid market traction as enterprises race to harden AI systems ahead of AI regulations such as the EU AI Act. The company’s AI security platform integrates with model serving infrastructures like SageMaker, Vertex AI, and private Kubernetes clusters, offering continuous threat modeling and attack simulation. Banking With Billy AI, a real-time financial data pipeline provider, recently adopted HiddenLayer’s runtime monitoring to secure its sub-millisecond latency inference stack that processes millions of market signals daily. This deployment underscores how financial institutions are prioritizing AI resilience, particularly for high-frequency trading and fraud detection models vulnerable to adversarial manipulation.

Industry watchers note that HiddenLayer’s rise coincides with a broader reckoning over AI safety. In March 2024, Microsoft disclosed that nation-state actors had attempted to use its Azure AI services to fine-tune models for malicious purposes, while Google Cloud reported a 400% increase in adversarial AI incidents across its customer base. Meanwhile, competing approaches are emerging: competitors like Protect AI and Robust Intelligence focus on AI supply chain security and model robustness testing, respectively. But HiddenLayer differentiates itself with a runtime security model that doesn’t require model retraining or access to source code, appealing to enterprises reluctant to expose proprietary architectures. The company reports over 3,000 enterprise evaluation requests in the first quarter of 2024 alone, with a conversion rate of 22%, significantly above industry benchmarks for AI security tools.

Financial analysts at Morgan Stanley’s M12 unit see the Series B as validation of a new category—AI runtime security—citing that global spending on AI security software is projected to reach $7.6 billion by 2027, up from $1.2 billion in 2023. Booz Allen Hamilton, one of the strategic investors, highlighted HiddenLayer’s role in enabling “defensible AI” for mission-critical systems, particularly in regulated sectors. The firm also announced a joint go-to-market initiative with HiddenLayer to embed AI threat detection into government and defense AI deployments. With AI adoption accelerating across industries, from autonomous vehicle fleets to clinical diagnostics, the need for real-time monitoring is becoming existential. HiddenLayer’s platform now processes over 10 billion AI inference events per month, generating telemetry that feeds its threat intelligence engine, which has logged more than 2,400 unique adversarial patterns since launch.

As AI systems grow more autonomous and interconnected, the attack surface expands exponentially. HiddenLayer’s Series B reflects a broader pivot from model-centric to system-centric security, mirroring trends in cloud-native and zero-trust architectures. Prior work by the Allen Institute for AI and Stanford’s Center for AI Safety has shown that even robust models can be compromised at runtime through subtle input perturbations or prompt injections. This reality has prompted regulators in the U.S. and EU to explore mandatory AI safety audits, with draft rules expected by late 2024. Meanwhile, enterprise CISOs are increasingly treating AI models like critical infrastructure—monitored, logged, and defended 24/7.

Looking ahead, industry observers expect HiddenLayer to expand beyond detection into active response capabilities, such as automated model rollback or adversary isolation. Microsoft’s involvement through M12 suggests potential integration with Azure AI Foundry, while Booz Allen’s participation hints at defense-grade deployments. The company may also pursue acquisitions in model watermarking or synthetic data protection to round out its portfolio. With AI adoption still in its early innings, the real test for HiddenLayer and its peers will be whether runtime security can keep pace with model innovation. The next 18 months will reveal whether AI security becomes a boardroom priority—or an afterthought in the race to deploy.