Google and YouTube face a dilemma on the intellectual protection of works created by AI

Google has decreed that the entire free internet is available to it for free to train its AI. But for Youtubethis poses real problems.

If a US court recently ruled that the intellectual property of works created by AI could not be protected, the rights holders of works used to train them would like to go much further. Starting with the Universal label, which includes many of the most listened to artists on YouTube, and that the streaming platform does not necessarily want to upset.

On the battery side, Google does not bother with intellectual protection to develop its AI

For almost all laboratories developing artificial intelligence, the current trend is towards the race for data. They are trying to get as much of it back as possible before the inevitable laws on the matter come to limit the possibility. As such, Google has not bothered with precautions and has simply appropriated the entire known Internet.

Of course, neither Google nor any other company currently developing AI pays the rights holders of the works from which they recover copies, nor anyone from whom they recover personal information. And faced with lawsuits which have already begun to multiply, they generally plead “fair use”, which authorizes the use of material protected by copyright under certain conditions assessed on a case-by-case basis by the courts and which gives the instant often right at these labs.

In other words, for the moment, Google believes that from the moment it is to train an AI, everything is accessible to it, and the company is not liable for anything. A policy that could well cause problems for its YouTube platform.

On the face side, rights holders are essential for YouTube

Last April, a track called Heart on my sleeve, sung by Drake and The Weeknd, is a great success. The only problem is that neither of them actually sang to this music, and it’s actually AI-generated vocals. A case that we have already seen more than once. Universal, the label of the two artists, did not like this at all and asked the platforms to remove the track. While some of them, like Spotify, performed quickly, YouTube didn’t. Google’s platform has a little less control over its catalog and relies on violating its content policies (like copyright infringement) to remove a video.

Except that the vocals cannot be eligible for copyright, and Universal managed to get around the problem by finding a sample itself protected in the song. But for all the actors present, this case announced a problem that should be settled concerning works created by artificial intelligence. Indeed, even if the law is not on its side, Universal is far too big for YouTube to afford to lose.

As so often with Universal, there’s no problem money can’t fix. This is how on August 21, YouTube announced that it was working with the label to create a system to remunerate the rights holders of music that was used to generate other music, even though nothing, in theory, is there. obliged. For this, the streaming platform is once again relying on its content ID, its highly controversial plagiarism detection robot. In short, YouTube is thus creating a parallel right to protect Universal’s dominant position on its platform, which will protect its own.

Sources: The Verge, Youtube