Have been working on a project which allows you to check if any of your source code has been copied from major online projects (over 70 billion lines of code) and report back if duplicates have been found. You can control the length of the run (so 15 lines means a match or 7 lines) and make the line matching fuzzy.

The idea being to provide compliance checks such as licensing and such, although I guess it would allow for checking for potential source code leaks as well and as a plagiarism detector.

It works by running a program over your source code which produces line by line hashes, with the result uploaded and then scanned over. The result is a report indicating what percentage of code was identified as copied both on a project and on an individual file level, with the identified source and its location. As such if nothing is leaked there is no ability to reconstruct the source code as it is a hash.

It works over 200+ languages, and while working over line by line currently can be expanded out to work in different ways.

Looking for interested parties to try it out (see my bio for contact) in addition to the current ones I have. Even if its just you wanting to kick the tires.