Meta Halts Plans To Train AI Models Using EU Users’ Data After Pushback From Lead Regulator

Meta will stop training its AI systems using data belonging to EU users, the tech giant confirmed today.

The decision follows a serious pushback from the EU’s top regulator body, the Irish DPC which is keeping strict tabs on how companies in this part of the world are functioning and collecting data.

The Irish DPC acts on behalf of a few data protection firms seen in this bloc. As per a press release by ICO, Meta has also been asked to justify all security and safety concerns that many regulators in this part of the world may be having with its AI systems.

The DPC mentioned how it does welcome such decisions across the board where Meta would be pausing plans to ensure training of LLMs gets fulfilled. Remember, both Facebook and Instagram have a lot of users, and having their material shared and then used for training is not something appreciated by the DPC.

The move followed an intense debate and meeting between the tech giant the DPC who did not allow Meta to go forward with its training plans. Meta hoped to tap into user-based content to conduct AI training so that markets across the US, EU, and beyond can benefit from modern AI with up-to-date material. But now, there is a huge obstacle that has come in its direction.

Meta has long argued how it was also looking to better its AI system as well as its LLMs through training done via user-generated content.

But everything seemed to be going downhill for Meta in May this year when the company sent out an alert to users about changes made to its privacy policy. And one of the biggest alarming signals had to do with how its apps had the right to make use of all content posted publicly.

This meant images, captions, interactions, comments, and whatnot would be included. Even a status published publicly would be used. How’s that for a rude awakening?

Meta further went on to justify this move by stating how it had to do this to better reflect different languages, cultures, and geographies across those situated in the EU.

The changes were all said to come into play by June of this year but now, it looks like that is not happening by miles. Europe’s GDPR is a tough law to follow and permission needs to be taken first before moving forward with any tech giant’s decision.

Meta kept on defending itself with a clause that mentioned it had all sorts of legitimate interests for its actions. It even delineated how the move was in line with the law. But such a legal basis being a part of its justification is not something new.

We’ve seen Facebook’s parent firm speak a lot more in this regard in the past and it’s been called out for doing so. The company has gone to great extents to make sure users cannot opt out of data tracking.

Close to two billion notifications were rolled out where users were informed about the new changes. They came in the form of a standard alert, similar to how you get for birthdays and friend requests. Hence, the strategy employed was evident in terms of how Meta just wanted fewer users to notice what exactly was taking place here.

Even if you did, the chances of you knowing how to opt out of this would be at a bare minimum, experts added. If you did take out time, you’d notice how there was a right-to-object form attached. Again, why Meta is making things so hard for people is beyond our minds.


Image: DIW-Aigen

Read next: Snowden Condemns OpenAI's NSA Hire, Privacy Concerns Erupt
Previous Post Next Post