Apple Clarifies: YouTube Resources Used for AI Training, Not Applied to Apple Intelligence

Apple Addresses Controversy Over Using YouTube Subtitles for AI Training

On July 18, a report revealed that tech giants like Apple and Nvidia had utilized YouTube subtitles to train their artificial intelligence models, sparking widespread discussion. Apple has since responded, clarifying the intent and nature of its data usage.

The company confirmed that it used YouTube datasets to train its open-source OpenELM model, which was released in April of this year. However, Apple stressed that OpenELM is not employed to power any of its AI or machine learning functionalities, including the "Apple Intelligence" service. The model was primarily developed to advance technological research and foster the growth of open-source large language models. Apple researchers have referred to OpenELM as "a cutting-edge open-source language model."

According to Apple's statement, the development of OpenELM was strictly for research purposes, and it is published as open source, making it widely accessible to researchers. Because OpenELM is separate from Apple's AI services, this means that YouTube subtitle data has not been utilized in the training of those services.

Additionally, Apple noted that its AI service models are trained on "licensed data," which includes curated data selected to enhance specific functionalities and publicly available data collected via its web crawlers. The company also indicated that there are currently no plans to develop a new version of the OpenELM model.

Most people like

Find AI tools in YBX

Related Articles
Refresh Articles