Meta has announced it will be training its AI models using public content shared across its platforms, including Facebook and Instagram, by UK adults, after it had paused training in the UK and the EU following data protection concerns.

Public posts across its main social media platforms will be used to train its upcoming and current AI models, including photos, captions, and comments.

Training content will not include private messages or information from accounts of users that are under 18 years of age.

The announcements marks a u-turn of Meta’s decision to pause the launch of AI models, and their training, in Europe after the Irish data protection regulator told Meta to halt its plans to using social media posts for training.

Meta has said that, since then, they have engaged with the UK’s Information Commissioner’s Office (ICO) and has welcomed the watchdog’s “constructive approach.”

“This clarity and certainty will help us bring AI at Meta products to the UK much sooner,” Meta said in their announcement.

According to the ICO, Meta has made changes to its Ai training policy, which includes making it easier for users to object to the processing of their data for AI training, as well as widening the window to do so.

The ICO says it will continue to monitor the situation as Meta presses forward with their plan.

“We have been clear that any organisation using its users’ information to train generative AI models needs to be transparent about how people’s data is being used,” Stephen Almond, executive director of regulatory risk at the ICO, said.

“Organisations should put effective safeguards in place before they start using personal data for model training, including providing a clear and simple route for users to object to the processing.”

He also warned that the ICO has not approved the processing outright, and that Meta must continue to ensure and demonstrate it is compliant with UK privacy laws.

From this week, UK-based adult users of Facebook and Instagram will begin to receive in-app notifications explaining Meta’s new AI training scheme, including how users can object to their data being used to train the company’s generative AI models.

Recommended reading

• Demystifying GDPR & AI: Safeguarding Personal Data in the Age of LLMs
• Meta Delays AI Training in Europe Following Regulatory Concern
• X’s Default Data Harvesting for AI Model Under Regulatory Scrutiny

The investigation into Google’s AI training practices follows a request from the Irish DPC for X to stop training its AI chatbot, Grok, on user posts until its GDPR compliance has been examined.

For those that have already objected to their data being used fro AI training, Meta will not contact and their choice will be honoured, the company said.

The objection form is said to be simpler than previously, due to feedback from the ICO.

Back in June, Meta’s AI training policy came under fire from privacy watchdogs across Europe and data privacy advocacy group NOYB. Currently, Meta has made no announcement saying it will resume AI training in the EU.

In somewhat of a protest to Europe’s relatively strict data privacy laws, Meta CEO Mark Zuckerberg and Spotify chief Daniel Ek penned a letter warning that Europe risks falling behind due to its complex regulatory landscape.

And this regulatory landscape has teeth: the EU is opening an investigation into Google’s AI training methods, saying that the tech giant may not have acquired prior consent to the processing of user data to train its generative AI models.

The investigation follows a request from the Irish Data Protection Commission for X to stop training its AI chatbot, Grok, on user posts until its GDPR compliance has been examined.

While different companies developing AI may have taken disjointed methods to train their models, the EU is staying steadfast in its protection of data privacy.

It remains to be seen if the ICO will find Meta’s updated training policy compliant with UK data protection law as the company steams ahead.