▼ Summary
– Meta will use public data from European users on Facebook and Instagram to train its generative AI models, notifying users in the EU and UK about this change.
– The data for training includes publicly shared posts, photos, captions, comments, and interactions with Meta’s AI chatbots, but excludes private messages and applies only to users aged 18 and over.
– Users will receive notifications within apps and via email, with links to an objection form allowing them to opt out of having their public data used for AI training.
– Meta argues that using regionally specific data is crucial for developing AI that understands local context, dialects, cultural references, and humor, enhancing the relevance and accuracy of its AI features for European users.
– The move reflects ongoing tensions between tech companies’ data needs and EU data protection regulations, with Meta’s opt-out mechanism being a key response to regulatory feedback.
Meta is moving forward with plans to use public data from its European users on platforms like Facebook and Instagram to train its generative artificial intelligence models. The company confirmed it will begin notifying users in the European Union and the UK this week about how their information will contribute to developing AI features.
This move revisits territory Meta stepped back from last year following pushback from privacy regulators, particularly Ireland’s Data Protection Commission (DPC), the lead regulator for Meta in the EU.
What Data is on the Table?
According to Meta, the data used for training will include publicly shared content. Think posts, photos (and their captions), and comments made publicly on Facebook and Instagram. It also includes interactions users might have had with Meta’s own AI chatbots.
Crucially, Meta emphasizes that private messages exchanged between friends and family will not be used for training these general AI models. The company also stated this data usage applies only to users aged 18 and over.
User Notification and the Right to Object
Users in the EU and UK should expect notifications within their apps (Facebook, Instagram) and via email detailing these changes. These notifications will reportedly contain a link to an objection form, allowing users to opt out of having their public data used for this specific AI training purpose. This opt-out mechanism appears to be a key component added in response to earlier regulatory feedback concerning user control under GDPR (General Data Protection Regulation) and similar UK data laws.
Even without the direct notification, users should eventually find access to this objection form within Meta’s updated privacy policy documentation.
Meta’s Rationale: Building Regionally Relevant AI
Why does Meta want this data? The company argues that training its AI models – including those powering features like Llama – on regionally specific data is essential for creating tools that understand local context. This includes grasping different dialects, colloquialisms, cultural references, and even distinct forms of humor and sarcasm prevalent across various European countries.
Meta suggests this is particularly important for multimodal AI, which deals with text, images, video, and voice, requiring a nuanced understanding of how these elements are used locally. The goal, Meta claims, is to make its AI features more accurate and relevant for European users.
Context: Data Appetites and Regulatory Hurdles
This isn’t entirely new ground. Meta announced similar plans for UK users previously and acknowledged last year it had already trained models on a massive dataset of public Facebook text and photos posted by adult users globally since 2007.
The renewed push in Europe highlights the ongoing tension between tech companies’ need for vast amounts of data to build sophisticated AI and the robust data protection frameworks established in regions like the EU. While Meta is providing an opt-out, the effectiveness and accessibility of this mechanism will likely be watched closely by users and regulators alike. The company is betting that the notification process and opt-out provision will satisfy regulatory concerns that previously put these plans on hold.
(Inspired by: The Verge)
