Conclusion of preliminary investigation X formerly Twitter: use of personal data for training the AI Grok
The FDPIC concludes his preliminary investigation into the use of personal data for training the artificial intelligence (AI) Grok. Users of Platform X can object to their public posts being used to train AI.
The FDPIC contacted the operator of Platform X, Twitter International Unlimited Company (TIUC), as part of an informal preliminary investigation after it became known in early summer 2024 that TIUC was using data from Platform X users to train the AI Grok. In this context, it was clarified how transparent this data processing was and what opt-out options were available.
During the preliminary investigation, the company named a representative in Switzerland and provided specific information on the processing of public posts for the purpose of training machine learning and artificial intelligence models, including generative models. In particular, TIUC informed the FDPIC about the opt-out option introduced since 16 July 2024. This allows users to reject the default use of their X contributions for training and fine-tuning Grok in the data protection settings. The FDPIC concluded that the company is complying with the requirements of the FADP by offering this opt-out option, which is also offered in the EU.
When data is used to train Grok, not only TIUC but also the users of Platform X are responsible for making use of the design options provided for organising the use of their own data.