Apple refutes claims of using YouTube videos for AI training
Apple has officially denied using its open-source OpenELM model to power any artificial intelligence (AI) or machine learning features, including Apple Intelligence. The tech giant's statement comes in the wake of an investigation that revealed major tech companies, including Apple, were utilizing YouTube content to train their AI models without the permission of creators. The dataset reportedly used encompassed over 170,000 videos from popular YouTube channels such as MKBHD and MrBeast.
OpenELM model solely for research purposes
Apple has clarified to 9to5Mac that the OpenELM model, released in April, was created exclusively for research purposes. The company aimed to contribute to the research community by advancing open-source large language model development. Apple researchers have previously described OpenELM as a "state-of-the-art open language model." The model is published open-source and is widely available, including on Apple's Machine Learning Research website.
Apple Intelligence unaffected by 'YouTube subtitles' dataset
Apple further clarified that the "YouTube subtitles" dataset does not power Apple Intelligence, as OpenELM is not used in its development. The company has previously stated that its intelligence models were trained "on licensed data, including data selected to enhance specific features, as well as publicly available data collected by our web-crawler." Additionally, Apple revealed it has no plans to build any new versions of the OpenELM model.
