Meta announced on Friday that it has paused its plans to train its large language model using public content shared by adults on Facebook and Instagram in the EU/EEA. This decision follows a request from the privacy regulator of Ireland, where the company has its international operations center. The Irish Data Protection Commission (DPC) welcomed Meta’s decision to pause its plans and described its conversations with Meta as “intense.”
The DPC had previously raised concerns about Meta’s AI training practices, particularly in relation to data privacy. Meta was set to expand its AI to Europe on June 26 but had to postpone those plans in order to comply with the requests of the DPC. Despite this, Meta expressed disappointment at the DPC’s decision and stated that without including local information, they would only be able to offer people a second-class experience.
While Meta’s plans for Meta AI in Europe are currently on hold, the company will collaborate with the DPC and address specific requests it has received from the UK Information Commissioner’s Office before resuming training. The activist group NOYB, which had pressured European data protection authorities to stop Meta’s plans, celebrated the decision and criticized Meta’s communication methods regarding the objections process.
Last year, the DPC fined Meta a record $1.3 billion for violating privacy rules. This history of regulatory issues has led to heightened scrutiny of Meta’s AI training practices, particularly in the European Union.
