There is increasing resistance against AI companies that use copyrighted data without permission to train their models. A startup in the music sector is now even openly admitting this. It’s all ‘fair use’, they say.
CEOs of several websites recently spoke out and strongly criticized the AI startup Anthropic for its behavior. In order to train its artificial intelligence, the company massively accessed data from the sites and the numerous requests almost brought the servers of iFixit and Freelancer.com to their knees. In addition, Reddit has just banned Microsoft from using information from the platform for AI training purposes.
Now there is new trouble, like Engadget reported. This time with the companies Suno and Udio, which develop AI in the field of music. Their software is supposed to make it possible for anyone to compose music, even if they have no idea about notes, harmonies and instruments. But in order to provide a functioning product, Suno and Udio’s models also have to be trained with masses of data. And this is done quite unashamedly and again under questionable circumstances.
Last month, publishers such as Warner, Sony and Universal sued the two companies. The accusation was that copyrighted material was being scraped and used without permission to train the AI. Instead of denying this accusation, however, Suno is now going on the offensive.
It is no secret that among the tens of millions of recordings used to train Suno’s model, there are probably recordings whose rights belong to the plaintiffs in this case. [Die Trainingsdaten] contain essentially all music files of reasonable quality that are available on the open Internet. Suno
However, the company does not see any problem with this situation. It is ‘fair use’ if a copy of a protected work is used as part of a technical backend process to create a new product that does not itself infringe copyright. The scraping that was carried out is therefore not illegal. The Recording Industry Association of America (RIAA), which co-filed the lawsuit against Suno and Udio, sees things somewhat differently, as expected.
Your infringement on an industrial scale cannot be called ‘fair use.’ It is not fair to steal an artist’s life’s work, extract its core value and repackage it to directly compete with the originals. RIAA
As is often the case with new technologies, legislation lags far behind everyday reality. In America, the ruling in the case against Suno and Udio is likely to set a precedent that will have a major impact on future rulings. Since the legal system in Germany is not based on common law or case law, in which rulings from the past are also decisive, it will probably take significantly longer in this country until legal questions surrounding AI training and the data used for it are clarified.
Alexia is the author at Research Snipers covering all technology news including Google, Apple, Android, Xiaomi, Huawei, Samsung News, and More.
