Meta Training its AI Chatbot On User Data
In order to train its artificial intelligence (AI) models in the UK, Meta had intended to begin gathering publicly accessible user data and information communicated with its chatbot, Meta AI, on June 26. The Irish Data Protection Commission (DPC), however, ordered the social media behemoth to delay its plans in order to allow the organization adequate time to resolve the users' privacy concerns. While the UK and other European countries have institutionalized such data protection procedures, most other parts of the world do not.
Highlights:
- Meta paused UK user data collection for AI training due to privacy concerns raised by European regulators.
- This delay applies to Europe due to stricter data protection laws.
- Meta still collects similar data elsewhere for AI training and regular operations.
For example, Meta has been gathering data from its US consumers since last year on all of its platforms and beyond. The internet behemoth also gathers this kind of information from consumers in many other nations, including India. This implies that whatever information a user has volunteered or made publicly available to Meta AI has probably already been utilized to train its Llama AI models and personalize advertisements.
Meta's privacy policy states that the platform gathers a lot of information about users based on how they use its products and apps. Among them are any publicly accessible posts, comments, audio, images and their description, watched or interacted-with advertisements, applications and features utilized, hashtags used, in-app purchases and transaction history, and the amount of time, frequency, and duration spent on an app that is owned by Meta. various apps are available on various platforms and include Facebook, Instagram, WhatsApp, and Meta AI.
Most applications gather some information, such as the features utilized, transactions made, and advertisements watched, in order to enhance the user experience, personalize advertisements, or improve the product itself. But some of the information gathered by Meta swiftly enters the morally dubious territory. For instance, the massive social media company also gathers user information regarding off-products.
"We collect and receive data from our partners, measurement vendors, marketing vendors, and other third parties about many different kinds of your data and actions on and off our Products," the business said in its privacy policy.
This data includes, but is not limited to, device information, websites visited, applications utilized, off-product sales and transactions, demographic information, and more. Meta has previously provided an explanation of how it uses user data to train its AI models in a blog post.
Given the DPC's recent move, Meta is probably going to postpone gathering data in the UK in order to train its AI. The General Data Protection Regulation (GDPR) and other privacy-related rules in Europe have caused unexpected delays in the rollout of Meta AI across the area. The massive social media platform will still gather data for its regular operations, though.