Meta’s new, advanced AI models cannot be used in the EU. The parent company of WhatsApp and Facebook has decided not to release its AI models in the European Union, as EU regulators provide too little clarity about training AI with data from European users.
In June, it emerged that Apple would not be releasing its new AI features, called Apple Intelligence, in Europe due to concerns about EU regulations. Meta is the latest tech giant to decide not to offer certain new AI products in the EU. “We will be launching a new multimodal AI model in the coming months, but not in the EU, due to the unpredictable way European regulators deal with AI,” Meta told Axios.
Multimodal AI models can handle text, video, audio and images. Meta plans to use the new multimodal Llama 3 model in its own apps and Meta Ray-ban smart glasses. In addition, other companies can use the model under an open license.
Meta says it will prevent companies in the EU from offering products and services that use the new models. A text-only version of Llama 3 will be released later, which will also be available in the EU.
Do not train AI with EU user data
Meta is currently banned from training its AI models using data from European users, as this would violate EU privacy law GDPR. The company announced in May that it wanted to use all public posts on Facebook and Instagram to train future AI models. But EU regulators put a stop to that, prompting Meta to suspend the plan.
Training with data from Europeans is crucial to ensure that new AI features “well reflect the terminology and culture of the region,” according to Meta.
Meta now complains to Axios that the EU regulators are working very slowly. The social media company also criticizes that competitors such as Google and OpenAI have been training their AIs with European data for a long time and that they have not yet been reprimanded by the EU.
Learn more about AI and stay up to date with our newsletter.
Source: www.bright.nl
