Meta will withhold its multimodal AI models from the EU amid regulatory uncertainty. This decision could prevent services using Meta’s AI models from being offered in Europe.
Meta announced that it will not offer its upcoming multimodal AI model and future versions to customers in the European Union due to regulatory ambiguity. These models are designed to process text, images, and audio, enhancing AI capabilities in Meta platforms and Ray-Ban smart glasses.
Meta stated to Axios, “We will release a multimodal Llama model over the coming months, but not in the EU due to the unpredictable nature of the European regulatory environment.”
Meta’s Concerns with GDPR Compliance
Despite this, Meta plans to release Llama 3, its upcoming text-only model, in the EU, but their primary concerns revolve around the challenges of training AI models using data from European customers while adhering to the General Data Protection Regulation (GDPR).
In May, Meta announced plans to use publicly available posts from Facebook and Instagram to train future AI models but had to stop in the EU after pushback from data privacy regulators. Meta defended its actions, emphasizing the importance of training models on European data to accurately reflect local culture and terminology.
Meta stated in a blog post, “If we don’t train our models on the public content that Europeans share on our services and others, such as public posts or comments, then models and the AI features they power won’t accurately understand important regional languages, cultures or trending topics on social media. We believe that Europeans will be ill-served by AI models that are not informed by Europe’s rich cultural, social and historical contributions.”
Broader Regulatory Concerns
Meta’s decision to abandon the launch of Llama 3 in the EU comes amid broader concerns about the Digital Markets Act (DMA) and the upcoming EU AI Act. The DMA aims to prevent anti-competitive behaviors, but many companies, including Meta, view the regulatory environment as unfavorable. Compliance with GDPR rules has already compelled Meta to halt training its AI models using posts from Facebook and Instagram users due to potential privacy violations.
The EU recently issued compliance deadlines for AI companies to adhere to the upcoming AI Act, requiring changes in areas such as copyright, transparency, and AI output by August 2026. Meta’s concerns mirror those of Apple, which also decided against launching its Apple Intelligence features in Europe due to the DMA. Apple cited potential security risks and challenges in meeting interoperability requirements.
Plans for the UK
Despite reservations about releasing multimodal models in the EU, Meta plans to launch them in the UK, which has similar data protection laws to the EU. The company argued that European regulators are slower in interpreting existing laws compared to their counterparts in other regions.
Featured Image courtesy of CFOTO/Future Publishing via Getty Images
Follow us for more updates on Meta and other AI news stories.