Elon Musk’s X under fire for allegedly using EU user data to train AI chatbot Grok without consent

Elon Musk’s X is facing scrutiny from the EU over the collection of users’ data to train its artificial intelligence chatbot Grok without consent. Ireland’s Data Protection Commission (DPC) is now taking legal action against X for violating the bloc’s privacy regulations.

According to a report from Irish broadcaster RTE, the DPC has initiated High Court proceedings due to worries about the use of European users’ data for the next version of Grok, which Musk has hinted will be released this month. Graham Doyle, spokesperson for the DPC, stated that the commission had been in discussions with X for several months, most recently last Thursday, and was surprised by these developments.

“The DPC says it concerns centre around the use of this data, in public posts by X users, in European Union/European Economic Area to train artificial intelligence systems utilised by Twitter including its enhanced search tool known as ‘Grok’,” RTE reported.

The DPC contacted X today and is expecting further engagement early next week. Kevin Schawinski, CEO of a Swiss AI company, criticized the platform, saying, “X added a setting for ‘we’ll take your data to train Grok’ without any notice and defaulted to ‘yes’ for everyone. This is BAD.”

In July, X made a change that automatically activated a setting for all users, allowing the use of their public posts to train its AI chatbot. The DPC told TechCrunch that it was surprised by X’s decision, as it had been in contact with the company about this issue for months. Although X has provided a help page since at least May instructing users on how to opt-out, it did not inform them that the default setting was to allow data access.

The DPC acknowledged that X had given users the option to opt-out but argued that this was insufficient. A significant number of European X users had their data processed without the protection of these measures. According to the DPC, X’s use of data to train Grok breaches the EU’s General Data Protection Regulation (GDPR) by not offering a timely opt-out mechanism.

The timing of when X began gathering users’ posts remains unclear. Currently, the platform operates on the assumption that users consent to the use of their posts and interactions for training Grok.

Launched last year, xAI aims to be a “maximum truth-seeking AI.” The rollout of Grok for Premium+ subscribers showed Musk’s ambition to shake up the AI field, as he discussed with Lex Fridman, expressing disappointment in OpenAI’s shift away from its nonprofit, open-source origins.

Musk first introduced Grok on November 6, 2023, describing it as having a “rebellious streak” and a sense of humor and sarcasm, distinguishing it from competitors like ChatGPT and Google Bard. The company said the first version of the chatbot was not trained on X data, and in March, it announced an update, though it remains unclear which data was used for training.