Apple is shifting its data policy to allow the use of user interaction data for training its artificial intelligence models. The company previously maintained a strict policy against using personal customer information for model development, focusing instead on on-device processing and private cloud computing.
The new approach involves using anonymized or authorized data to improve machine learning systems. Apple plans to implement a controlled, phased mechanism where users will have the option to opt-in to share de-identified text inputs, Siri interaction logs, and application feedback.
To protect user privacy, the company will continue to utilize differential privacy and local synthetic data generation techniques. These methods are designed to prevent the association of sensitive personal information with model weights during the training process.
This strategic change addresses the growing need for high-quality, real-world interaction data. Industry observers note that the lack of real-time user feedback has hindered Apple's ability to compete with rivals like OpenAI, Google, and Meta in developing advanced large language models.
