Post #3148391

The concept is simple: decompose a program into atoms, identify the unique mal-ecule that makes up the program, and use a fast local ML to keep the false-positive rates low. Treat binaries and source code as first-class citizens, with automated reverse-engineering of both sets. We still have a long way to go, but if you want to check out where we are today (including a web portal for analyzing samples), see https://atomdrift.org/