September 26, 2023
McLean, VA - ĢƵ Allen Hamilton (NYSE: BAH)—the largest single provider of artificial intelligence services for the federal government—today announced that its corporate venture capital arm, ĢƵ Allen Ventures, has made a strategic investment in , a security platform that safeguards machine learning (ML) models. This investment strengthens and expands ĢƵ Allen’s tenured adversarial AI capabilities, including those developed for the Department of Defense and intelligence clients, and will further accelerate secure adoption of enterprise AI solutions to keep pace with emerging national security threats as well as rising consumer expectations.
“Every AI-enabled solution should be assessed for risk and appropriately protected from adversarial attacks—especially as the government looks to deploy AI capabilities in increasingly important applications,” said Matt Keating, leader of ĢƵ Allen’s Adversarial AI portfolio. “Our clients operate in complex environments that require AI models to be highly specialized, rapidly deployable, and secure. The HiddenLayer investment by ĢƵ Allen Ventures better positions us to integrate startup, commercial, and open-source innovation to rapidly augment our existing capabilities. Ultimately, it allows us to more quickly and confidently deliver robust AI capability to our clients—and the country at large.”
With increased AI adoption, specifically AI models deployed within mission-critical systems, the risk surface increases for federal, defense, civil, national security, and commercial users, with bad actors looking to exploit and accelerate cyber threats. The newly announced investment in HiddenLayer will complement and accelerate ĢƵ Allen’s existing adversarial AI capabilities, a leader for over five years in advancing machine learning (ML) methodologies to safeguard systems against attack. This includes a long-standing focus on addressing key challenges with model security, such as data poisoning, data leakage, model evasion, and malicious code injection. In addition, ĢƵ Allen has also led advanced research to assess the adversarial image perturbation robustness for computer vision models and how manipulated tabular data can enhance the behavior evasive capabilities of Microsoft Windows malware.
“Using pre-trained open-source models is an overall net positive, but this foundation also puts AI models at greater risk for adversarial attack. This tension is a threat that organizations need to be aware of, plan for, and get ahead of, as our adversaries are doing just that,” said Edward Raff, chief scientist at ĢƵ Allen and leader of the ĢƵ Allen ML research team, which has been publishing academic research on adversarial AI since 2018.
This is the latest AI-focused investment by ĢƵ Allen Ventures, which identifies and invests in strategic, dual-use commercial technologies, with recent investments including Shift5,Credo AI,Hidden Level,Latent AI,Synthetaic, and Reveal Technology. The investment also builds on ĢƵ Allen’s focused efforts and missions around generative AI and responsible AI, providing a robust security foundation as AI use increases.
“HiddenLayer’s powerful platform and expert team has proven effective in securing AI from a broad range of threats, so we quickly identified them as a partner that can support and protect our AI deployments,” said Travis Bales, managing director at ĢƵ Allen Ventures. “From our early discussions, it was clear to us that the HiddenLayer team has the vision and execution to continue developing security for the emerging AI market.”
ĢƵ Allen’s recent investment now enables federal agencies to capitalize on HiddenLayer’s award-winning , as well as ĢƵ Allen’s AI security research, risk and vulnerability assessments, and managed detection and response services—all paired with AI security engineering best practices, tools, and technologies.
"ĢƵ Allen continuously proves its commitment to developing AI capabilities that are robust, secure, and offer the technical depth needed by the federal government. Their 360-degree approach to AI combined with their steadfast commitment to HiddenLayer's vision since our founding made them a perfect partner for the next stage of our growth,” said Chris Sestito, co-founder and CEO of HiddenLayer. “Bringing together our MLSec platform and their purpose-built AI solutions ensures the government can continue to innovate through AI adoption with confidence knowing they are secure from all types of cyberattacks, including those from nation-states."
Learn more about ĢƵ Allen’s work in adversarial AI and about ĢƵ Allen Ventures.