The terms of art are Automatic Content Recognition https://en.wikipedia.org/wiki/Automatic_content_recognition and Video Fingerprinting https://en.wikipedia.org/wiki/Digital_video_fingerprinting

You can dive in from there but it's basically either watermarking or fingerprinting of video and or audio frames. Video was preferred because there were fewer false positives from music beds. In a nutshell its video Shazam