"Computer Science Researchers Lead the Defense against the Dark Side of Tech"

There is a dark side to Machine Learning (ML) and Natural Language Processing (NLP) advancements, as they eliminate the need for malicious actors to employ human workers to duplicate and distribute false or malicious content. According to Virginia Tech Associate Professor of Computer Science Bimal Viswanath, malicious actors can now use these models for malicious campaigns. Spammers, advertisers, attackers, and malicious actors can now control how language models generate text with much higher variability without developing complex algorithms. Bots can be instructed to speak like a poet or as members of certain demographic groups, making it easy to weaponize text and more challenging to identify it as artificial. Internet users who consume enormous amounts of information are susceptible to manipulation. Even the most critically literate users will likely be unable to distinguish between reality and falsely created information. In an effort to disarm weaponized media and toxic misinformation campaigns, Viswanath and his research team have been working on methods for detecting deepfakes. Focusing on proactive detection and protection techniques, the group continues to meet efficacy challenges. As a result of the rapid development of deepfake technology, the landscape consists of sophisticated algorithms engaged in a constant game of cat and mouse. The attacking algorithm changes to outmaneuver the defender when a credible defense is established. The research team led by Viswanath accumulated a large data set of Artificial Intelligence (AI)-manipulated media. This data set, which more than 60 research groups use worldwide, enables researchers to examine firsthand how their detection systems perform outside the lab against adaptively responsive attacking algorithms. This contribution to future detection and security efforts bridges the gap between the laboratory and real-world application. This article continues to discuss the team's efforts toward improving the detection and disarming of weaponized media and toxic misinformation campaigns.

Virginia Tech reports "Computer Science Researchers Lead the Defense against the Dark Side of Tech"