Research, health care, and science all keep racing ahead, and human research protection programs keep trying to keep up. Studies involving development and testing of machine learning/artificial intelligence projects that use software were just not on most institutional review boards’ radar 10-15 years ago, but now they’re here, and we are learning how to best review them. The software at the heart of ML/AI projects is a device and these studies may fall under Food and Drug Administration oversight. This Public Responsibility in Medicine and Research blog post talks about some of the considerations related to determining whether ML/AI research is subject to FDA regulations. Of particular interest is the link to an FDA “Discussion Paper and Request for Feedback,” which, while not final, will give some ideas of important considerations. (It’s 20 single-spaced pages long, so it may be a while before we’re able to consider all those considerations.)