Elon Musk Halts Use of European X Posts for AI Training—For Now
Elon Musk has temporarily agreed to stop using posts from European X users to train his Grok chatbot, according to a recent announcement.
Ireland’s Data Protection Commission (DPC), which oversees privacy for X under the General Data Protection Regulation (GDPR), released a statement on Thursday welcoming the decision to "suspend processing of personal data for the purpose of training the AI tool Grok."
“The DPC commends X’s agreement to halt the processing of personal data from posts made by EU/EEA users between May 7, 2024, and August 1, 2024, intended for training its AI Grok,” the regulator stated in an emailed press release. “This decision came amid urgent court proceedings initiated by the DPC under Section 134 of the Data Protection Act, 2018, before Ms. Justice Reynolds, who emphasized the importance of data subjects' rights across the EU/EEA in her remarks.”
Earlier this week, the DPC initiated court proceedings seeking an injunction against X for processing data without user consent. The matter has also been reported by Ireland’s national broadcaster, RTE, which plans to refer it to the European Data Protection Board (EDPB).
In response to X's announcement, DPC commissioner Des Hogan remarked, “My colleague, commissioner Dale Sunderland, and I are pleased with X’s decision to suspend processing while we, alongside our EU/EEA counterparts, assess compliance with the GDPR. One of our key roles as an independent regulator is to safeguard data subjects’ rights. Today's decision reinforces our commitment to protecting the rights of X users across the EU and EEA.”
The DPC has received inquiries regarding whether it will pursue the deletion of any unlawfully processed data. Questions remain about the legality of AI models trained on such data and how privacy authorities will address these issues.
OpenAI’s competitor, ChatGPT, has also drawn scrutiny from GDPR enforcers, particularly regarding the processing of European public data for training. In January, Italy's privacy watchdog issued warnings to OpenAI over potential GDPR violations.
Despite an EDPB taskforce's report in May addressing GDPR application to ChatGPT, key issues regarding the lawfulness and fairness of data processing remained unresolved.
Demanding the deletion of models trained on unlawfully sourced data would represent a significant move for privacy regulators.
Update: The DPC has provided an update: assistant principal communications officer Risteard Byrne stated, “The matter is adjourned and will return to court in September.” He confirmed that the DPC will continue to evaluate X's processing actions with EU/EEA colleagues for GDPR compliance.
X also responded to inquiries, sharing a link to its Global Government Affairs account, which mentioned: “We are pleased that EU users can continue using Grok while controlling their data through a simple privacy setting. We will maintain collaboration with the DPC regarding Grok and other AI-related matters.”
In light of X's statement, we sought confirmation from the DPC regarding a legally binding assurance that X has ceased processing EU user data for AI training. Byrne indicated that the agreement was reached in the Irish High Court, led by Ms. Justice Reynolds but provided no further details.
The High Court's media relations office could not locate documentation related to the agreement when reviewed. Media relations advisor Gerry Curran noted, “It’s likely the undertaking was made verbally in court, as we cannot find any filed paperwork.”
In an earlier post, X criticized the DPC’s injunction efforts, calling them “unwarranted, overbroad, and unjustified.” The company asserted, “While many firms scrape the web for AI training data at the expense of user privacy, X strives to empower users with better control over their data.”
Ranking
