Meta has received approval from the European regulator to train its artificial intelligence (AI) models on publicly available social media content.
Tech and social media giant Meta has announced its intention to begin training proprietary AI tools using publicly available content, such as public posts and comments shared by adults on Meta-owned platforms within the European Union.
In addition, interactions with Meta AI, including user questions and queries, will be utilised to help train and enhance the given models.
The company owns and operates Facebook, Instagram, Threads, and WhatsApp, among other products and services. It invests heavily in the development of artificial intelligence. This year alone, the corporate management plans to spend about $65 billion on AI-related projects. One of its latest developments is a large language model, Llama 4, which is expected to be released later this month after a few delays.
The new AI training initiative builds on the successful launch of Meta AI in Europe last month. Meta AI is Meta’s personalised virtual assistant designed to integrate seamlessly across its platforms. It supports six European languages: English, French, German, Italian, Spanish, and Portuguese. Users in 41 European countries and 21 overseas territories can interact with the virtual helper to brainstorm ideas, plan trips, or get answers to questions using information from the web. It is accessible within Meta’s major platforms through a new blue circle icon. The tool’s features are currently limited to text-based interactions, with functionalities like image generation and editing not yet available due to regulatory considerations.
Regulatory concerns have also been the reason why Meta delayed the use of public user content to train its large language models last year. Now, legal and regulatory clarity has been established. The company received the green light from European data regulators.
In December 2024, the European Data Protection Board (EDPB) issued an opinion confirming that Meta’s original approach to using public content aligned with legal obligations under EU data protection laws. Following that, the tech giant engaged in further discussions with the Irish Data Protection Commission (IDPC). These constructive consultations helped ensure that Meta could proceed with its plans legally and responsibly.
Today, the number of Meta AI’s active users is reportedly over 700 million per month. The company claims to better support millions of individuals and businesses across the EU by incorporating region-specific content into their AI services. Leveraging original local user posts and queries, Meta’s AI systems might more accurately understand and reflect the diverse cultures, languages, and needs of the European communities.
The use of data for AI purposes shall be voluntary. Meta Platforms users across the region will soon start receiving notifications that explain how and for what their data may be utilised. These messages will include a link to an opt-out form, where they can submit their objections anytime.
According to the company’s announcement, private messages sent via Meta services do not fall under the category of data used for AI training. This concerns only the public content, which other users can also see on social media, or the history of communication with AI assistants.
