News
TRAILS Researchers Part of Team Receiving $1.8M DARPA Award to Make AI More Trustworthy
UMD’s Soheil Feizi and Furong Huang are collaborating with experts at NYU to make AI-driven large language models more consistent, adaptable and secure in high-stakes environments.
UMD Researchers Investigate Security Threats to Web AI Agents
In a first of its kind study, UMD’s Furong Huang uncovered crucial factors underlying vulnerabilities of web AI agents, highlighting the need for enhanced security measures.
Feizi Receives $1M Award to Advance the Foundations of Reasoning AI Models
UMD’s Soheil Feizi is a recipient of the Presidential Early Career Award for Scientists and Engineers (PECASE).
Anti-Smoking Chatbots Provide Sound Advice—Most of the Time
GW’s Lorien Abroms and David Broniatowski evaluated three chatbots’ responses to users who asked for information about ways to stop smoking.
TRAILS Leadership Part of Winning Team in UMD Invention of the Year Competition
Hal Daumé III and Katie Shilton were part of a group that won in the social innovation category for their technology that can teach students in any discipline about AI and information literacy.
To Make Language Models Work Better, Researchers Sidestep Language
UMD's Tom Goldstein was featured in Quanta Magazine for developing a novel language model that reasons in latent space and learns when to switch to language, significantly improving efficiency and performance on complex tasks.
Shittu Explains How AI Can Drive Informed Decisions on Energy Use and Sustainability
In a new video, GW’s Ekundayo Shittu discusses applications of AI used in systems engineering and the energy sector.
How Images Reflect AI’s Dark ‘Spiral’
UMD’s Cody Buntain shows how generative AI outputs can skew negative, even after positive prompting.
TRAILSCon 2025 Examined Trustworthy AI at Work
Experts in industry, policy and academia convened at a two-day conference hosted by TRAILS in February at the George Washington University.
Decoding Emojis with Artificial Intelligence
UMD’s Wei Ai is testing AI’s ability to interpret emojis in various scenarios and languages, with the goal of improving their use in machine learning and natural language processing models.
Harnessing Market Forces: UMD Team Incentivizes AI Companies to Prioritize Safety
UMD’s Furong Huang has developed the first auction-based AI regulation system that encourages companies to compete on safety compliance instead of capability.
UMD Researchers Build AI Database to Improve Math Learning Outcomes
Supported by a $4.5 million grant from philanthropic organizations, Jing Liu and Wei Ai are creating an open-source, high-quality dataset of classroom recordings that can be used to accelerate AI-driven outcomes for K–12 math education.
Autonomous Cars Don’t Understand How Blind People Move Around. A UMD Research Team Is Trying to Boost Safety.
Hernisa Kacorri is developing a dataset with real-world 3D motion-capture data and detailed descriptions that accurately capture what blind individuals encounter in an urban setting.
UMD Researchers Organize NeurIPS Competition to Improve AI Watermarks
UMD's Furong Huang and Tom Goldstein led an international AI watermark-removal competition to test the resilience of watermarking techniques for identifying AI-generated images.
Feizi Receives Presidential Early Career Award for Scientists and Engineers
UMD's Soheil Feizi has been honored with the Presidential Early Career Award for Scientists and Engineers, the U.S. government's highest recognition for outstanding scientists and engineers early in their careers.
College Students Struggle To Balance AI Use With Academic Integrity Policies Set By Universities
MSU’s Virginia Byrne told the University Herald that she encourages students to evaluate AI's strengths and weaknesses, promoting ethical practices and trust as colleges navigate the integration of AI into education and career readiness.
College Students ‘Cautiously Curious’ About AI
MSU’s Virginia Byrne studies the role of technology in education, with a focus on how it impacts college students. In an interview with the States Newsroom, Byrne provides insight on how colleges should be handling AI.
UMD Team Uses AI to Enhance Mnemonic Learning
UMD's Jordan Boyd-Graber is leading the development of an AI-driven mnemonic generator designed to simplify vocabulary learning.
UMIACS Team Aims to Boost High-Performance Computing Software Development Using AI
UMD’s Tom Goldstein is leading a $7 million DOE-funded project to enhance AI-assisted software development for high-performance computing, aiming to improve the effectiveness of large language models in parallel programming and boost developer productivity.
Leveraging Natural Language Processing to Guide Educators
UMD’s Jing Liu is using natural language processing to analyze K-12 math classroom transcripts and provide feedback to help teachers improve, finding that NLP is more effective when combined with human coaching.