This week's report Wired It was revealed that Apple, NVIDIA, and other tech companies are using hundreds of thousands of YouTube videos to train their AI. The YouTube captioning dataset contained transcripts of videos from educational and online learning channels ranging from MIT and Harvard to MrBeast, MKBHD, and PewDiePie.
After the controversy, Apple 9to5Mac The company's open source OpenELM model reportedly uses the dataset, although Cupertino claims the model is not used for any AI, machine learning, or Apple Intelligence features.
According to the publication, Apple created the OpenELM model “as a means to contribute to the research community and advance open source large-scale language model development (…) OpenELM was created for research purposes only and is not intended to be used to power Apple Intelligence features.”
In a previous statement, Apple said the underlying models are trained on “licensed data” and don't use users' private, personal data or interactions.
Why is the Apple Intelligence Beta release being delayed?
While Apple claims that it doesn't use any stolen or personal user content in its AI, the delay in releasing the official beta of Apple Intelligence is intriguing: the company had previously said it would release it for testing this summer alongside the iOS 18 public beta.
The public beta has been released, and Apple has issued new developer beta 3 builds, but Apple Intelligence is not available. Additionally, the blog post that said the company's AI would be released this summer alongside the public beta has been removed.
The release date is mentioned only as “later this fall” as a beta feature alongside iOS 18, iPadOS 18, and macOS Sequoia.
This doesn't mean that Apple delayed the AI release to remove references to stolen content, but it's interesting that Wired's report coincides with Cupertino removing references to the Apple Intelligence release.
For more details, please see below.
