What if an unstable aircraft approach could be identified early enough for a flight instructor to intervene? A developer built an open-source early-warning system that uses public ADS-B data from the OpenSky Network to flag unstable approaches by general aviation training aircraft near Daytona Beach International Airport (KDAB). In an initial three-hour dataset, 105 approaches were analyzed, 90 of them by training aircraft, with 72 selected for instructor review and 7 flagged by at least one rule; one Cessna 172S approach triggered both speed and sink-rate warnings. The project's next step is validating and calibrating its rules against expert human judgment, and it cautions that flags do not confirm a genuinely unstable approach given wind effects and ADS-B measurement limits. What if an unstable aircraft approach could be identified early enough for a flight instructor to intervene? That is the problem behind this open source project: AI Early Warning for Unstable Approaches in GA Flight Training. The project uses public ADS-B data from the OpenSky Network around Daytona Beach International Airport KDAB to analyze general aviation training approaches. It currently: The initial dataset covered 3 hours of flight activity: 105 approaches analyzed 90 training aircraft approaches 72 approaches selected for instructor review 7 approaches flagged by at least one rule One Cessna 172S approach triggered both speed and sink-rate warnings, highlighting how data-driven methods could support instructor review. Importantly, the project does not claim that every automated flag represents a genuinely unstable approach. Ground speed is affected by wind, ADS-B measurements have limitations, and instructor review is essential. That is what makes the project interesting: the next step is validating and calibrating the rules against expert human judgment. This is a good example of how open source aviation data can be turned into a practical safety research problem. 🔗 GitHub: https://github.com/samsuseelan/ga-unstable-approach-warning https://github.com/samsuseelan/ga-unstable-approach-warning