- Meta was planning to use EU users’ data, which is available publicly on Instagram and Facebook, to train its AI models.
- This would have helped the company understand its language, culture, and local trends and offer a personalized experience to users from that region.
- However, many European regulators have raised concerns against this process and asked Meta to hit the brakes on its plans until further review.
Meta AI was set to roll out a program where Meta would use the data of users from the European Union and the UK to train its AI models. However, this has now been forced to hit pause.
The company was set to roll out Meta AI, users had been notified about it and the changes were due to come into effect on June 26. But things took a bitter turn for the company.
It started with an advocacy group called NOYB filing a complaint against Meta AI to authorities from Austria, France, Belgium, Greece, Germany, Italy, Ireland, the Netherlands, Poland, Norway and Spain.
Here are the objections it raised:
- Instead of asking users explicit permission to use their data, Meta simply added the “opt-out” option.
- The notification was sent alongside regular Facebook notifications such as birthdays and new likes and comments. So there’s a high chance that users might miss it completely. And if they do, they won’t know that they can opt out of sharing their data.
- Opting out itself wasn’t as simple. The user had to fill out an objection form, explaining why they didn’t want their data processed and it was up to Meta to decide if it wanted to accept it.
The company even believed that its actions were completely legal and cited “legitimate interest” as a basis for it. But as it turns out, that’s not how it works.
The Irish Data Protection Commission (DPC), acting on behalf of all the other regulators, asked the company to delay its plans.
A similar objection has been raised by The U.K.’s Information Commissioner’s Office (ICO) as well, which asked Meta to hit the brakes until it can satisfy their concerns.
Although Meta has agreed to comply with the request of the regulators, it’s not very happy about this move. It even said that delaying AI training would be a huge setback for the EU and would cause further delays in bringing the benefits of AI to Europe.
The company is even more upset over the fact it had already discussed this project with the European DPAs in March and had incorporated the feedback they gave.
Meta was truly being cautious this time.
- It promised to only use publicly available information such as posts and comments, and steer clear from private chats.
- It also ensured that no data from users under 18 would be used in the process.
So being asked to pause the project after all this time and after so many precautions doesn’t make any sense to them.
Meta planned to use this data to offer EU citizens a personalized experience. This data would have helped it understand their local language, culture, and trends in a better way. In its absence, EU users would only get a “second-rate experience”.
The DPC on the other hand is very happy with Meta’s decision and said that it will continue to “engage” with the company until they find a mutually beneficial way out.
ICO executive director for regulatory risk, Stephen Almond, shared similar sentiments. He said that it’s great that Meta has addressed their concerns and is willing to pause the program until further review.
Almond also said these rules aren’t just for Meta. All major developers of generative AI will be subjected to similar types of monitoring.
He added that while generative AI can truly be transformative, it’s important that people can understand it and be assured that their privacy will be respected.
Max Schrems, chairman of NOYB, also welcomed this news but said the job isn’t done here. It will continue to track Meta’s activities.
“So far there is no official change in the Meta privacy policy, which would make this commitment legally binding. The cases we filed are ongoing and will need a determination,” he added.
It remains to be seen how this whole situation will pan out and whether the parties will be able to reach a middle ground.
Our Editorial Process