AI and Cyber: The Twain Have Met
Prof. Anupam Joshi, UMBC
Video Deepfake Classification Using Evolving Deep Neural Networks
Prof. Li Zhang, Royal Holloway, University of London
Beyond the Surface: How Post-Training Artifacts Shape LLM Diversity and Safety
Prof. Weiyan Shi, Northeastern University
UMBC is providing AI, machine learning, and cybersecurity experience, resources, and support to the University of Maryland, Baltimore (UMB) Institute for Clinical and Translational Research (ICTR). In turn, UMBC faculty are eligible for many of the ICTR's existing programs and resources for UMB researchers engaged in clinical translational research.
The UMBC expertise and systems will help UMB researchers explore the use of AI and machine learning to identify patterns in data and make decisions with minimal human intervention. For researchers developing applications and devices to improve the delivery and exchange of health information, UMBC's cybersecurity expertise can help protect those devices, systems, and data from attack. Read more about the ICTR programs and UMBC's collaboration here and here.
]]>A recording of the hybrid session of this event is now available on the NCCoE event page.
Cybersecurity and AI are two of the most important research areas for information systems today. Join the NIST National Cybersecurity Center of Excellence (NCCoE) on January 14th, 2026, for a hybrid workshop to discuss the Preliminary Draft of the NIST Cybersecurity Framework Profile for Artificial Intelligence (Cyber AI Profile).
Register now by Friday, January 9 to attend in-person (9-5) or virtually (9-12) to help shape the development of the NIST Cyber AI Profile.
Working with the cybersecurity and AI communities, the NIST NCCoE published a preliminary draft of the Cyber AI Profile to help organizations strategically adopt AI while addressing and prioritizing cybersecurity risks stemming from its advancements. The preliminary draft of the NIST Cyber AI Profile is available for public comment through January 30, 2026.
This workshop will feature in-person and virtual presentations in the morning, followed by in-person only breakout sessions in the afternoon. The breakout sessions will allow in-person attendees to engage directly with the NIST Cyber AI Profile team and for in-depth discussion of the preliminary draft of the Profile.
This publication is a preliminary draft – and your feedback is crucial in shaping the next and final version of this publication.
]]>AI is becoming more embedded in our daily lives - making everyday tasks easier and improving how we track and manage our health. But as these technologies grow more powerful, so do concerns about privacy and security. See a video of a talk by Ted Murphree, Manager of Security and Privacy Compliance at an international law firm and adjunct professor at UMBC-Shady Grove, and learn how everyday users can protect themselves from these risks.
Edward (Ted) Murphree is a seasoned cyber/privacy legal professional with over 20 years of experience in myriad legal and operational roles. Ted currently works as the Manager for Security and Privacy Compliance for a large international law firm, where he operationalizes risk management processes involving privacy, security, and AI.
In addition to his law degree, Ted holds a Ph.D. in Engineering Management with emphasis in information security and five information privacy certifications through ISACA, IAsecurity, and PP. Ted enjoys learning and has taught a course in Cybersecurity Law and Policy for UMBC-Shady Grove the last few years. His dual perspective in law and operations provides an additional layer of understanding to this continuously evolving field.
Joint work with Yuechun Gu and Jiajie He
To appear, 2015 Int. Conf. on Knowledge Discovery and Data Mining
With increasingly deployed deep neural networks in sensitive application domains, such as healthcare and security, it is essential to understand what kind of sensitive information can be inferred from these models. Most known model-targeted attacks assume attackers have learned the application domain or training data distribution to ensure successful attacks. Can removing the domain information from model APIs protect models from these attacks? Our work studies this critical problem. Unfortunately, even with minimal knowledge, i.e., accessing the model as an unnamed function without leaking the meaning of input and output, the proposed adaptive domain inference (ADI) attack can still successfully estimate relevant subsets of training data. We show that the extracted relevant data can significantly improve the performance of model-inversion attacks, for instance. Specifically, the ADI method uses the concept hierarchy extracted from the public and private datasets that the attacker can access, and it applies a novel algorithm to adaptively tune the likelihood of leaf concepts in the hierarchy showing up in the unseen training data. For comparison, we also designed a straightforward hypothesis-testing-based attack called LDI. Among all candidate methods, the ADI attack extracts partial training data at the concept level, converges fastest, and requires the fewest target-model accesses.
Dr. Keke Chen is an associate professor in the UMBC CSEE Department. His recent research focuses on privacy and security issues with AI model training and deployment. He earned his PhD in computer science from Georgia Tech in 2006. Before joining UMBC, he was a Northwestern Mutual associate professor of computer science at Marquette University.
Support for this event was provided in part by the NSF under SFS grant DGE-1753681.
Reinforcement learning (RL) has demonstrated remarkable success across diverse domains, from mastering complex games to optimizing real-time feedback systems in robotics and industrial control. However, its potential in cybersecurity, particularly for autonomous attack detection and mitigation in distributed systems, remains largely underexplored. Traditional single-agent RL approaches struggle in decentralized environments where multiple entities make independent decisions, necessitating multi-agent reinforcement learning (MARL). Our research explores blockchain networks as an ideal test case due to their decentralized architecture and trustless consensus mechanisms. We developed a novel MARL-based consensus mechanism for Proof-of-Stake blockchains, enabling nodes to collaboratively identify and penalize malicious behavior while preserving decentralization. This approach effectively mitigated six major blockchain attack types with minimal computational overhead. Building on these results, we propose integrating game-theoretic principles into the MARL framework to model adversarial strategies and enhance system resilience. The synergy between reinforcement learning and game theory establishes a robust foundation for dynamic and adaptive security in distributed systems, effectively addressing current vulnerabilities while anticipating and countering future threats. This integrated approach enables the design of resilient, scalable defense mechanisms tailored to the complex dynamics of decentralized architectures.
CLARK is a digital repository aimed at advancing cybersecurity education by delivering high-quality curriculum resources to educators throughout the United States. We will explore the design science methodology that guided its development, emphasizing how these principles maintain the repository’s usability and relevance for users. Additionally, the presentation will address the integration of artificial intelligence into CLARK and how its extensive dataset of curriculum content has facilitated this process. Attendees will learn how CLARK meets the changing demands of cybersecurity education and contributes to shaping the next generation of cyber professionals.
Dr. Blair Taylor is the Director of the Center for Interdisciplinary and Innovation Cybersecurity (Cyber4All) and a Professor in the Department of Computer and Information Sciences at Towson University. She is a national expert in cybersecurity education and curriculum development. She worked with NSA’s College of Cyber as a subject matter expert on long-term strategies to increase the pipeline of qualified students and build the nation’s cyber workforce.
Paige Zaleppa is a Ph.D. student in Information Technology at Towson University, where she is studying the intersection of cybersecurity education and artificial intelligence. She has led the development of the CLARK project since late 2019 and has been an adjunct faculty member at Towson University since 2021. She is passionate about connecting educators and students with educational resources that can be utilized to create the next generation of cyber heroes.
Support for this event was provided in part by NSF under SFS grant DGE-1753681. The UMBC Cyber Defense Lab meets biweekly, Fridays 12-1pm. All meetings are open to the public.