Apple’s OpenELM AI Model Was Reportedly Trained on YouTube Videos’ Subtitles.
Apple has reportedly stated that its recently released Open-source Efficient Language Models (OpenELM) AI model is not utilized for its Apple Intelligence features or any other machine learning functionalities within the company. This clarification from the Cupertino-based tech giant comes in response to concerns raised after an investigation revealed that the OpenELM AI model was trained on datasets that included subtitles from hundreds of thousands of YouTube videos. These same datasets were also reportedly used to train AI models from companies like Anthropic, Salesforce, Nvidia, and others.
In a conversation with 9to5Mac, Apple emphasized that OpenELM does not power any of the AI features branded under Apple Intelligence, nor does it play a role in other machine learning capabilities that Apple employs. The company clarified that OpenELM was developed as a contribution to the research community, with the aim of advancing the field of large language models. This distinction is significant as it underscores that Apple’s proprietary AI systems are independent of the open-source model it released earlier this year.
Apple further explained that the OpenELM AI model was specifically designed for research purposes and was not intended for integration into any of Apple’s in-house AI technologies. This statement aims to dispel any misconceptions about the role of OpenELM in Apple’s machine learning ecosystem. Importantly, OpenELM remains an open-source project, and its model weights are available to the public through Apple’s listing on Hugging Face, a popular platform for AI research and development.
The controversy surrounding OpenELM arose after reports suggested that the AI model had been trained using subtitles from a vast number of YouTube videos. This sparked concerns about the ethical implications of using such data, especially in light of the growing scrutiny on how AI models are trained and the potential biases they may inherit from their training data. Apple’s decision to clarify the role of OpenELM in its AI strategy appears to be an effort to distance itself from any negative perceptions associated with the model’s training data.
Despite the controversy, Apple’s move to open-source the OpenELM model has been viewed positively by many in the AI research community. By making the model freely available, Apple is contributing to the broader effort to advance AI research, enabling other researchers and developers to experiment with and build upon the model. This aligns with the company’s broader commitment to supporting open-source initiatives and fostering innovation in the tech industry.
In summary, Apple’s clarification that OpenELM is not used for its Apple Intelligence features or any other machine learning projects within the company serves to address concerns about the model’s training data and its potential impact on Apple’s AI systems. By positioning OpenELM as a research-focused project, Apple continues to support the development of AI technologies while maintaining the integrity of its own proprietary AI solutions.