Report from The Decoder
In Brief – A German regional court has cleared Meta to use public data from Facebook and Instagram users to train its AI systems, dismissing an emergency lawsuit from a consumer protection group. Back in April, Meta announced that it would start using publicly available information from adult users, including names, profile photos, comments, and likes, to train its AI large language models. The court ruled that Meta’s AI training plans fits within the framework of the General Data Protection Act (GDPR) as a legitimate corporate interest, as well as leaning on a 2024 determination of the European Data Protection Board (EDPB), which says companies can use publicly available adult data under data protection rules. The company swore to the court that sensitive details like account information, home addresses, or license plate numbers won’t make it into the training data. In addition, users can opt out of the use of their data for AI training. The decision of the German court aligns with the recent decision of the Irish Data Protection Commission (DPC), Meta’s lead privacy regulator, to approve the plan.
Context – Last fall, Meta spearheaded an open letter joined by over 50 companies, including EU tech leaders SAP, Spotify, and Ericsson, calling for “clear rules, consistently applied, enabling the use of European data” to build AI services for European citizens. Even with its comprehensive AI Act legislation and specialty EU AI Office, overlapping regulatory authorities, especially GDPR-enforcers, have created regulatory uncertainty in Europe. The EDPB’s report on Chat-GPT raised numerous questions about the valid legal basis for all stages of personal data processing, as well as raising the concern that hallucinations that are produced by all chatbots could violate the GDPR’s “principle of data accuracy”. The Irish DPC, often criticized for being too lenient on top US tech firms, has taken aggressive positions on AI training, leading Meta and X to stop training on data from European users as well as opening an investigation of Google practices.