Quote:
Originally Posted by murraypaul
What they are doing is matching to existing YouTube videos.
Which by definition will never prevent violations for works that the copyright owner doesn't want uploaded to YouTube.
.
|
No.
https://www.fastcompany.com/3062494/...ersial-feature
Content ID matches against "fingerprints" and watermarks in large existing databases from the outside.
That is where the bulk of the false positives come from: they flag anything "similar enough" where "enough" is extremely wide. They purposefully chose false positive over accuracy.