无忧传媒

Federal agencies are investing in AI to automate processes and enhance decision making across mission systems, ranging from critical national security assets to essential healthcare platforms. Yet, as these AI systems proliferate, they remain vulnerable to a growing set of attacks.

Such threats are quickly becoming a national security concern. 鈥淚nterest in deploying AI solutions in the government continues to climb, with use expected to grow 40% this year,鈥� explains 无忧传媒 Principal Matt Keating,聽leader of the firm鈥檚 AI security practice. 鈥淏ut, with a few notable exceptions, we haven鈥檛 seen commensurate energy devoted to developing robust AI models.鈥� This is not unique to the federal government. A recent Gartner survey revealed that of the organizations polled, only 4% are implementing AI application security tools.聽

无忧传媒鈥檚 AI security team is prepared to help. These experts have been working with federal agencies to understand the threat landscape and AI security for more than five years鈥攁nd, with increased AI adoption, they anticipate seeing more demand for their services.

Serving clients across the federal government in such a complex and evolving field demands expertise that鈥檚 both deep and wide. 无忧传媒鈥檚 AI security team delivers this on all levels.聽

Matt is an experienced business leader and technologist who has worked across industries鈥攊ncluding healthcare, law enforcement, and finance鈥攁nd areas of technical expertise, from bioinformatics to cryptography. Chief Scientist Edward Raff, who leads 无忧传媒鈥檚 machine learning (ML) research team, is a recognized expert in several ML subfields including malware detection, reproducibility, and ethics. Together, Matt and Edward have assembled a team with similarly rich and diverse backgrounds.

Lead Machine Learning Scientist Amol Khanna, for example, previously used machine learning to safely automate redundant tasks for a leading global bank. He also built a differentially private federated learning algorithm for disease prediction. Since joining 无忧传媒, he has performed research on safe ML methods essential for deploying models trained on sensitive data into real-world environments.聽 聽

Senior Lead Scientist Andre Nguyen started with 无忧传媒 as an intern and grew into聽solutions architecting and ML research roles, building expertise in areas including cloud data platforms, pharmacovigilance, adversarial machine learning, and cybersecurity research. He left the firm to oversee ML for a biotech company, then returned in 2023.聽聽

Lead Scientist Derek Everett has a background in computational high-energy nuclear physics. Since joining 无忧传媒, he has performed research targeting the safety of ML systems, as well as their application to malware detection.聽

鈥淚'm excited to advance the state-of-the-art for adversarial AI while increasing clients鈥� effectiveness in this space,鈥� Derek says.聽

In a market currently dominated by open-source tools, off-the-shelf software, and one-size-fits-all solutions, the approach of 无忧传媒鈥檚 AI security team is different by design.

鈥淲e're not simply reusing the same template each time,鈥� says Andre. 鈥淓ach algorithm and use case is different, often requiring specific solutions.鈥�

Edward explains one of the reasons why. Often there鈥檚 a big difference between adversarial AI in theory鈥攖he 鈥減ristine academic laboratory,鈥� where many current solutions have been tested鈥攁nd adversarial AI in the real world, with constraints and incomplete information.聽

鈥淲hen you add operational complexity, many of these tools don鈥檛 work as advertised,鈥� he says.聽

This is where 无忧传媒鈥檚 extensive experience with government agencies comes in, as does its holistic, big-picture approach.聽

鈥淲e鈥檒l typically start by creating a threat model to make sure we are focused on the most probable and damaging attack vectors for the environment,鈥� says Matt.

鈥淲e look at the entire structure and process,鈥� says Edward. 鈥淔or example, if adversaries can attack a certain part of the system, then the most effective solution may be to design around the problem as a whole, rather than engineer a patch for the vulnerable part.鈥�

In short, says Edward, 鈥溛抻谴�媒 is able to create reliable tools for real-world missions.鈥�

Creating these AI security solutions鈥攁nd accelerating their development and deployment鈥攈as been a highly collaborative enterprise. For example, as Edward works on adapting algorithms and software to scale AI training, Amol focuses on out-of-distribution detection and differential privacy, both critical components to deploying secure AI models that protect individual privacy. 鈥淲e鈥檙e enhancing our client鈥檚 AI capabilities by developing robust tools that can be safely applied to sensitive training data,鈥� Amol says. 鈥淭his allows us to build solutions that protect against inversion attacks.鈥�

鈥淥ur clients and the general public rightfully have anxiety about all the plethora of ways AI can fail,鈥� says Edward. 鈥淚鈥檓 excited to work with a team that is helping everyone, including our federal clients, build confidence in the function of the algorithms.鈥�

鈥淎s a team committed to enhancing the nation鈥檚 military readiness, intelligence, and citizen services, we are motivated not only to help our clients implement AI solutions, but also do so in a reliable and safe manner,鈥� says Matt.