Researchers have developed a technique to identify the origins of proprietary “black-box” AI models, even when their internal structure and training data are hidden.

Three Georgia Tech faculty members received Google Academic Research Awards to study how to make AI safer.