How to Limit Data Sharing with Meta AI Models
Meta, formerly known as Facebook, has recently come under scrutiny for its data sharing practices. The company’s privacy policy states that it uses “information shared on Meta’s Products and services,” such as posts, photos, and captions, to train its AI models. However, private messages are not used for training data. This practice has caught the attention of users in Europe, who received a notification about an update to Meta’s privacy policy as the company rolls out new generative AI features in the region.
Meta claims that it is notifying users in Europe in compliance with local privacy laws, particularly the General Data Protection Regulation (GDPR). UK-based user Philip Bloom received a notification stating that the changes would go into effect on June 26, 2024. On the other hand, users in the U.S. did not receive a notification, suggesting that the data sharing policy is already in effect for them.
Meta has been deploying generative AI features since September 2023. Initially, users were able to tag the Meta AI chatbot in conversations on Messenger, Instagram, and WhatsApp, and chat with AI personas based on licensed celebrity likenesses. More recently, Meta expanded its AI-powered features by making Meta AI the default search bar on Facebook, Instagram, Messenger, and WhatsApp, and allowing users to chat with the AI about specific posts. Unfortunately, these features cannot be turned off by users.
However, UK and EU-based users have the right to object and opt out of sharing their data to train Meta’s AI models. The process to opt out is tedious and intentionally designed to be difficult to navigate. Users have criticized Meta for making it challenging to exercise their rights. Nevertheless, opting out is possible through specific steps outlined by users who have successfully done so.
To truly cut off data from Meta, the only option is to delete your accounts. However, there are alternative methods for limiting data sharing. Meta provides an opt-out link, but it may only be available in certain regions. Another option is to submit a request to Meta’s help center, where you can access, delete, or file a complaint about personal information used to develop and improve AI at Meta. Notably, the form does not explicitly mention opting out of data sharing with the AI models.
Users are required to provide specific evidence when submitting a request, such as prompts that generated a response containing personal information or screenshots of the response. Meta reviews these requests based on local laws and may not fulfill them automatically. However, users in the EU or UK, where stricter privacy laws apply, may have an easier time accessing or deleting their data.
Another method for protecting your data from Meta is through the “activity off Meta” settings. This allows you to see sites and apps that share information with Meta. You can disconnect specific sites, clear previous activity, and manage future data sharing. Disconnecting and clearing previous activity will delete the data already collected by Meta. To prevent future data sharing with third parties, choose “Manage future activity” and select “Disconnect future activity.”
However, it remains unclear how these settings impact the data used to train Meta’s AI models internally. Meta’s privacy settings primarily focus on sharing with third parties rather than how Meta utilizes user data. Nevertheless, adjusting these settings is a step towards reclaiming privacy.
In conclusion, while Meta’s data sharing practices have raised concerns among users, there are methods to limit the amount of data shared with its AI models. Users can opt out, delete their accounts, or adjust their privacy settings to disconnect data sharing with third parties. However, it is essential to note that these measures may not completely address how Meta internally utilizes user data.
