
Meta announced today that it’s planning to resume the training of its AI models using public content from adult users in the UK. The company had started training its AI using public posts so it could better understand British culture, but the Facebook and Instagram owner had to hit the pause button to engage with regulators.
After receiving feedback from the UK’s Information Commissioner’s Office (ICO), Meta says that its approach to AI training is now more transparent. “We welcome the ICO’s guidance supporting Meta’s implementation of the legal basis of ‘Legitimate Interests’, which can be a valid legal basis for using certain first party data to train generative AI models for our AI at Meta features and experiences,” the company explained.
In the coming months, Meta will be using public content shared by adults on Facebook and Instagram in the UK to train its AI models. That will include posts, comments, and photos, including pictures from adults featuring minors. However, the company won’t be using private messages to train its AI. And adults in the UK will still be able to opt out.
“While our original approach was more transparent than our industry counterparts, we’ve incorporated feedback from the ICO to make our objection form even simpler, more prominent and easier to find,” the company said today. “We’ll honour all objection forms already received, as well as new objection forms submitted.”
Meta AI is currently available in 22 countries around the world, and the intelligent assistant works across WhatsApp, Instagram, and Facebook. In the US, it’s also on Ray-Ban Meta smart glasses and Meta Quest headsets.
Earlier this week, Melinda Claybaugh, Meta’s global privacy director made headlines after admitting that the company had been using posts from Australian users from as far back as 2007 to train its AI models. And Meta won’t allow adults in Australia using its products to opt out of the AI training due to less restrictive local privacy laws.