A recent investigation by Proof News sparked controversy, alleging AI models belonging to some major tech companies, including Apple, Nvidia, Salesforce, and Anthropic were trained using a dataset containing YouTube subtitles.
The dataset called "YouTube Subtitles" made by non-profit EleutherAI contained subtitles from over 170,000 channels, the publication reported, including popular creators like Mr. Beast who has hundreds of millions of subscribers. There were allegations that tech companies were trying to profit from the data and the fire also came upon Apple"s OpenELM models, which used this dataset.
Apple came forward and confirmed to 9to5Mac that the OpenELM model isn"t used in Apple Intelligence or its other AI/machine learning features. In other words, the YouTube Subtitles database doesn"t power Apple Intelligence features.
OpenELM is a family of open-source models released earlier this year. The company describes OpenELM as "a state-of-the-art open language model" and says it was released to "empower and strengthen the open research community, paving the way for future open research endeavors."
OpenELM is accessible through various sources, including Apple"s Machine Learning Research website. However, Apple told 9to5Mac that it doesn"t have plans to build future versions of the OpenELM model.
The company previously clarified that it doesn"t use "users" private personal data or user interactions when training our foundation models" to train Apple Intelligence models. However, it does use "licensed data" and data collected by its crawler unless websites specifically tell the company not to do it.
We train our foundation models on licensed data, including data selected to enhance specific features, as well as publicly available data collected by our web-crawler, AppleBot. Web publishers have the option to opt out of the use of their web content for Apple Intelligence training with a data usage control.
Apple Intelligence is a suite of AI features baked into the upcoming iOS 18, iPadOS 18, and macOS Sequoia, scheduled to arrive later this year. The company will start testing Apple Intelligence features in beta this fall in U.S. English.
Source: 9to5Mac