Meta's AI Training with EU Data Sparks Privacy Issues

Meta’s use of data from European Union (EU) social networks users to train its artificial intelligence (AI) model aims to  improve algorithm accuracy. However, this practice has raised privacy and consent concerns under EU laws. 

A significant challenge Meta faces in training AI with user data is adhering with the European Union's stringent  privacy and data protection laws. The General Data Protection Regulation (GDPR) mandates a high standard for consent and data use, requiring companies to obtain explicit consent from individuals to process their data, especially for non-essential purposes.

Since May 22, Meta has sent 2 billion notifications and emails to European users, explaining its plans and providing a link to an online opt-out form, according to ABC. This new privacy policy is set to take effect on Jun. 26, with AI model training expected to commence shortly thereafter.

However, privacy advocacy groups such as None of Your Business (NOYB) argue that merely notifying users is insufficient. They contend that explicit consent must be obtained to comply with GDPR requirements. The organization highlights the ongoing debate over whether using personal data to train AI, even if it is publicly shared information, deviates from the original purpose users intended when posting their content.

"Shifting responsibility to the user is completely absurd. The law requires Meta to obtain opt-in consent, not to provide a hidden and misleading opt-out form,” stated Max Schrems, founder, NOYB. "If Meta wants to use your data, they have to ask for your permission. Instead, they make users beg to be excluded. We were especially surprised that Meta went to the trouble of incorporating tons of little distractions to ensure that only a small number of users would bother to object."

NOYB, has requested an urgent procedure from the data protection authorities of 11 European countries, including France, Germany, Italy, Ireland, Spain, among others, concerned about Meta Platforms' processing of data.  The request arises in the face of Meta's imminent implementation of AI without offering users the option to opt-out, raising concerns about GDPR compliance.

Users across the European Union who publish posts on platforms like Instagram and Facebook will implicitly contribute to training materials for Meta's larger language models. Meta's main purpose in using EU social network data to train its AI is to enhance the accuracy and effectiveness of its algorithms and machine learning models. This aims to enable them to offer more personalized and relevant services to users, such as “better language translation, more effective content moderation and an overall enriched experience on their platforms”.

A senior Meta executive, according to Blog.PL, announced that public Facebook and Instagram posts serve as a source of knowledge to enhance its Large Language Model Meta AI (LLaMA) models, while consciously excluding private and shared communications with friends to protect user confidentiality.

“With the creation of this technology comes the responsibility to develop best practices and policies that comply with local laws and regulations,” reads the press release. “We are consulting with our lead privacy regulator in the EU, the Irish Data Protection Commission, and have incorporated their feedback to date to ensure that the way we train AI in Meta complies with EU privacy laws.”