Skip to content

HiddenLayer

AI security platform unifying ML supply-chain scanning, runtime defense, posture management and automated red-teaming for enterprise AI systems.

Visit Website ↗
72/100Meaningful Innovator

Overview

HiddenLayer’s AISec Platform covers the AI asset lifecycle: scanning model files and ML supply-chain artifacts for tampering or embedded malicious code, detecting adversarial inputs and model-extraction attempts at runtime, mapping AI posture across an organization’s model inventory, and running automated red-team simulations against deployed models and LLM applications. The company positions itself around defending against prompt injection, adversarial manipulation, model theft and supply-chain compromise specifically for machine learning artifacts, which differ structurally from traditional software packages.

Founded in March 2022 by Tanner Burns and Chris Sestito, HiddenLayer is based in Austin, Texas, and employs roughly 169 people. It raised a $50M Series A in 2023 led by M12 (Microsoft’s venture fund) and Moore Strategic Ventures, bringing total funding to about $56M — a notable signal given Microsoft’s own venture arm backed a company competing in adjacent territory to Microsoft’s internal AI security tooling.

Innovation Matrix Assessment

Innovation Velocity 8/10

Built out supply-chain scanning, runtime defense, posture management and red-teaming as a unified platform within roughly three years of founding.

Operational Value 7/10

Model-file scanning and runtime detection integrate into existing ML pipelines, giving practical coverage of a concrete AI supply-chain risk (malicious pickle/model files).

Market Momentum 7/10

$50M Series A led by Microsoft's M12 fund is a credible strategic-investor signal, and headcount (~169) suggests real commercial traction, though no later round has been disclosed.

Category Disruption 7/10

Model-file and ML-artifact scanning addresses an attack surface (malicious serialized model files, model extraction) that conventional AppSec/SCA tools do not cover.

Real-World Efficacy 6/10

HiddenLayer's research team has publicly disclosed real model-security vulnerabilities and technique write-ups, giving more independent grounding than most peers, though platform-wide detection rates remain unverified.

Enduring Relevance 8/10

ML supply-chain and model-file risk is structural to how organizations consume open-source and third-party models, a trend that is accelerating, not fading.

Why CISOs Should Care

Closes a blind spot conventional software composition analysis misses entirely: malicious or tampered model files and ML supply-chain artifacts moving through MLOps pipelines.

What Makes It Different

Treats models and ML artifacts as a distinct asset class requiring their own scanning and runtime-defense layer, rather than extending generic file or network security tools to cover them.

The Matrix Verdict

72/100 — MEANINGFUL INNOVATOR

Strong Performer: a broad, genuinely AI-native platform with credible strategic backing and real vulnerability-research output, though it hasn't yet published independent, third-party-verified efficacy data at scale.

Editorial Note: Claims vs. Verified Findings

Employee count (~169) and Series A terms are corroborated across multiple data providers. Specific detection-rate or catch-rate claims made in HiddenLayer marketing materials are UNVERIFIED and were not independently tested for this profile.

Sources