Google claims YouTube terms allow AI music training rights
Google has asked a US federal court to dismiss a lawsuit brought by independent musicians over the use of copyrighted music to train its AI music model. The company argues that content uploaded to YouTube falls under a broad licensing agreement that authorises reuse by Google and its affiliates. The case centers on whether platform terms of service can override copyright claims in AI training disputes.
In its filing, Google says users who upload music to YouTube grant a worldwide, non exclusive and sublicensable licence that allows reproduction and creation of derivative works. The company relies on this clause to argue that training data used for its AI systems falls within permitted use. Google also states that its affiliates include Google DeepMind, the unit responsible for developing the AI music model known as Lyria 3.
The lawsuit was filed by a group of independent artists and producers who allege that Google copied tens of millions of audio clips from YouTube. They claim this material was used to train Lyria 3, which generates music tracks and vocal content for integration into Google’s broader AI ecosystem. The plaintiffs argue that the company moved from content distribution to direct competition by using creators’ work without compensation.
The dispute highlights growing legal uncertainty around AI training and copyright law. Google’s argument avoids the traditional fair use debate and instead focuses on contractual licensing through platform terms. Courts in the United States have recently issued mixed rulings on AI training, including decisions suggesting that non identical model training may qualify as transformative use, while other cases involving major tech firms continue to test the limits of copyright protection in machine learning systems.
-
23:00
-
22:45
-
22:15
-
21:45
-
21:21
-
19:30
-
19:15
-
18:05
-
16:50
-
15:29
-
14:14
-
13:45
-
13:12
-
13:05
-
12:45
-
12:15
-
11:25
-
11:04
-
10:59
-
10:19
-
09:59
-
09:44
