They claim that their model, trained with YouTube videos, is for research purposes only.
With the spread of artificial intelligence, companies interested in technology will have to make up for their shortcomings in the near future. In its current state, the legal background – although there are governing laws and regulations worldwide – cannot fully regulate the process and resources of training models.
Contemporary authors are suing AI companies one after the other for the unauthorized use of their content, and although no truly authoritative decision has yet been reached in the matter, it is only a matter of time when a precedent-setting ruling will make the obvious practice of the companies involved impossible. However, since the training of AI models is an extremely data-intensive process, for lack of a better one – despite a lot of side talk and obfuscation – most people use this method.
The latest company in the crosshairs is still new in the field, although it is one of the largest. Although Apple is significantly behind compared to other companies, it is preparing to make up for the disadvantage by publishing its own platform in the near future.
The AI development of the market-leading tech company under the name Apple Intelligence is also attacked with the claim that the company uses copyrighted content without permission or compensation, but of course the company denies this. According to their claim, they only used a database containing the transcripts of many YouTube videos for research purposes.
The open source model was trained with, among other things, the text version of 173 thousand videos from 48 thousand channels. According to a report published a few days ago, all this was done without permission. The database in question was created by a non-profit company called EleutherAI, and other companies besides Apple, such as NVIDIA, have also used it for a similar purpose.
By the way, according to its own claim, the apple company did not use the database directly for Apple Intelligence training, but for its open-source OpenELM models, which were released in April, according to their claim, only for research purposes. They stated that the new AI platform appearing in the 18th version of iOS debuting this year will not rely on it during its operation, and no new version will be made of it.
Source: www.pcwplus.hu
