Meta, the parent company of Facebook and Instagram, has announced plans to commence training its AI models using public posts from UK adults on its social media platforms.

The American technology firm indicated that this initiative will begin in the upcoming months following constructive discussions with the UK's data protection authority, the Information Commissioner’s Office (ICO).

Previously, the deployment of AI tools at Meta in the UK and EU had been stalled due to regulatory concerns, leading to a stalemate between the tech giant and EU regulators.

However, Meta expressed that it has had a positive engagement with the ICO, which has now approved the initiative.

In a blog post, Meta stated, “This clarity and certainty will help us bring AI at Meta products to the UK much sooner.”

The company welcomed the ICO’s guidance, which supports the use of ‘Legitimate Interests’ as a legal basis for utilizing certain first-party data to train generative AI models for its AI at Meta features and experiences.

Importantly, Meta clarified that it will not utilize private messages exchanged between users and their friends or family for training its AI models, nor will it use data from accounts belonging to individuals under 18 in the UK.

The firm plans to leverage publicly available information—such as public posts, comments, photos, and captions—from adult users on Instagram and Facebook to enhance its generative AI models.

Starting next week, users in the UK will receive in-app notifications detailing the initiative and instructions on how to access an objection form, allowing them to opt out of having their data used for AI training.

Meta also noted that it has taken the ICO's feedback into account to make the objection form “simpler, more prominent, and easier to locate.”

Stephen Almond, the ICO's executive director of regulatory risk, stated that the office will “monitor the situation” in the weeks ahead.

He stated, “We have made it clear that any organization utilizing user data to train generative AI models must be transparent regarding the usage of individuals' information.

“Organizations are required to implement robust safeguards prior to employing personal data for model training, which includes offering a straightforward and accessible option for users to refuse data processing. The ICO has not granted regulatory approval for this processing, and it is the responsibility of Meta to ensure and demonstrate continuous compliance.”

Former Deputy Prime Minister Sir Nick Clegg, who now serves as the president of global affairs at the technology company, expressed that the organization is “pleased” to initiate the program in both the UK and Brazil.

In a message posted on X, previously known as Twitter, he remarked: “This is a positive development for innovation in both nations – it allows us to introduce our AI products at Meta to the UK and Brazil much sooner, ensuring that our generative AI models are attuned to local culture, history, and language nuances.

“Regrettably, our efforts to train our AI models to appreciate the EU’s diverse cultural, social, and historical contributions are currently on hold as EU regulators have yet to reach a consensus on the application of the law.”