Meta has announced plans to withhold its upcoming multi-modal AI models from users in the European Union due to what it’s calling a lack of clarity from regulators. The company plans on releasing a larger, text-based version of its Llama AI model that will be made available for customers and companies in the region.
In a statement to Axios, the social media giant claimed that it 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.” The move shows an increased willingness of American tech giants to withhold products from their European customers. Apple also announced that it won’t be releasing its Apple Intelligence features in Europe because of regulatory concerns.
The decision will also keep European companies from being able to use Meta’s models, despite them launching under an open license. This in turn can also keep companies outside the EU from offering similar services and products to EU customers.
The real issue here isn’t the “unpredictable nature of the European regulatory environment,” but with GDPR, the EU’s current data protection law instead. While the AI Act is still being finalized, companies can’t train their AI models on data from customers in the EU without explicit consent. Meta, and other US tech companies, argue that this is hurting both EU customers and the competitiveness of European companies.
A Meta representative told Axios that training on European data is necessary for their models to “properly reflect the terminology and culture of the region.” The company also alleges that Google and OpenAI, the two biggest AI giants at the moment are already training their models on European data.
Regardless, Apple and Meta seem to think that withholding products and services from EU customers can pressure the union to relax their regulation around data handling for AI models, and eventually give them more freedom.
In the News: WazirX faces major security breach; $230 million stolen
