EU: Meta won't train its AI model with user data - for now
Meta won’t be using any data from users in the EU for machine learning purposes, at least not for now. The company made the announcement last Friday afternoon. In postponing its plans, the company is complying with an order from the Irish data protection authority, its lead regulator.
In the past weeks Meta had been informing users of Facebook, Instagram and Threads of upcoming changes to these services’ privacy policies. The changes were meant to go into effect on June 26.
The company is currently developing a system it calls Meta AI, which is meant to be able to generate text and images, a technology known as “generative artificial intelligence.” In order to teach the system language skills, the software is “trained” with vast quantities of data – like text written by human users. According to Meta, examples include “public posts, public comments, or public photos and their captions.”
Meta had no intention to ask for consent
The plans had drawn sharp criticism, owing to the fact that Meta did not intend to obtain express consent from users. In order to engage in this form of data processing, the company is required to have a legal basis in accordance with the General Data Protection Regulation (GDPR). Meta planned to claim a “legitimate interest.” The company did not plan to ask for consent; it only meant to allow users to object to the use of their data and opt out.
The Austrian data privacy advocacy group Noyb considers this measure insufficient, however. Earlier this month, the organization filed complaints with eleven national data protection authorities: in Austria, Belgium, France, Germany, Greece, Ireland, Italy, the Netherlands, Norway, Poland and Spain. In its complaint, Noyb alleged additional GDPR violations. Meta is attempting to prevent those impacted by its policies from exercising their right to opt out, Noyb alleges. “This change is particularly worrying because it involves the personal data of about 4 billion Meta users.”
Verbraucherzentrale NRW, the consumer advisory body for the German state of North Rhine-Westphalia, also criticized Meta’s plans, finding that the opt-out process was very cumbersome and not user friendly. Last week the consumer watchdog issued a warning to the company.
Data protection officials deliberate
Top Hamburg data protection official Thomas Fuchs stated last week that European data protection authorities were currently weighing whether Meta could claim a “legitimate interest.” His office was in contact with the lead regulatory authority in Ireland, where Meta has its European headquarters.
The Irish authority, the Data Protection Commission (DPC), has issued a statement, saying that it “welcomes the decision by Meta to pause its plans.” The DPC characterized Meta’s decision as the result of “intensive engagement” between the company and the authority. According to the statement, the DPC “will continue to engage with Meta on this issue.”
The company meanwhile said it was “disappointed” by the authority’s demand that it delay the use of data from users to train its algorithms. In a blog post, Meta claimed that the European data protection authorities have been informed of its plans since March. The company still believes that “our approach complies with European laws and regulations.” For now, Meta AI will not be launched in Europe. Meta also defended its actions, saying that other companies also made use of user data to train their AI models. Meta’s approach is “more transparent than many of our industry counterparts,” the company wrote.
Criticism from Noyb
Max Schrems, chair of Noyb, said in a statement: “We welcome this development but will monitor it closely. So far, there has been no official change to the Meta privacy policy that would make this commitment legally binding. The cases we have filed are ongoing and will require an official decision.”
Noyb’s was also critical of Meta’s announcement that European users for the time being could not use its AI system. The GDPR allows for almost any manner of data processing, Noyb wrote, as long as users choose to opt in – Meta could simply ask users for their consent.
Said Schrems: “The Meta press release reads a bit like ‘collective punishment.’ If one European insists on his or her rights, the whole continent will not get our shiny new products. But Meta has every opportunity to deploy AI based on valid consent – it just chooses not to do so.” (dpa / js)