According to a journalistic investigation led by Wired and Proof News, several YouTube content creators have found their clips stolen without their consent to train AI models.
What is it about, exactly?
Apple, Nvidia, Anthropic and other major market players are at the center of a major scandal. Subtitles from more than 170,000 videos belonging to popular names such as Marques Brownlee, MrBeast and Jimmy Kimmel were allegedly used by them after EleutherAI downloaded them. EleutherAI is a company that helps large companies train artificial intelligence models. The purpose of this approach? To create materials for academics and various small developers.
To avoid being finger-pointed, the dataset the firm created after downloading the subtitles contained only the plain text and translations into German, Japanese and Arabic. According to Wired, no representative from EleutherAI wanted to comment on the matter.
Apple and reactions from content creators
None of the YouTube creators consented to their videos being used to train AI models, nor were they even asked beforehand. One of them even deleted all his videos from the platform. In vain: they still exist in some of the AI models.
Marques Brownlee himself, visibly disturbed by the situation, posted his thoughts on X and created an Instagram reel. He concluded that “Apple and a lot of other tech companies train their AI models using data they buy from third-party companies, some of which obtain their data in slightly illegal ways. Apple can say they did nothing wrong” because they didn’t handle this whole process. A classic case of “blaming others” to avoid repercussions.
The whole situation brought to the fore by Wired and Proof News raises serious questions about ethics and copyright in the digital age. The rapid increase in the use of AI models trained on data taken without the permission of the creators shows the need for clear regulations and international consensus on the responsible use of digital data.
Source: Phone Arena
Source: www.go4it.ro
