#AI horizons 25-04 – META and EU

Meta Resumes AI Training in Europe as EU Advances AI Continent Action Plan

Meta’s Strategic Pivot on European AI Training Data

Meta has announced it will restart training its artificial intelligence models using publicly available content from European users, a process it had paused last year following privacy concerns. This resumption comes after European privacy regulators in December affirmed that Meta’s original approach complied with legal obligations under the EU’s data protection framework. According to Meta’s official blog post, “Making AI Work Harder for Europeans”, this training will “better support millions of people and businesses in Europe, by teaching our generative AI models to better understand and reflect their cultures, languages and history.”

The company plans to utilize public posts and comments shared by adult EU users on Facebook and Instagram, along with queries directed to Meta AI, to train and improve its models. As reported by AP News, Meta emphasized that private messages between friends and family will not be used for AI training, nor will public data from users under 18 years of age.

Beginning in April 2025, Meta started notifying EU users about this data usage through in-app and email communications. These notifications include a link to an objection form where users can opt out of having their data used for AI training, with Meta pledging to “honor all objection forms” the company has already received or will receive in the future.

The company has defended its approach by noting that competitors like Google and OpenAI have already used data from European users to train their AI models. According to The Hacker News, Meta positions this training as essential for ensuring its AI systems better understand and reflect European cultures, languages, and regional nuances, particularly as AI models become more advanced with multi-modal functionality spanning text, voice, video, and imagery.

This development marks an important shift in Meta’s European AI strategy and follows the company’s recent launch of Meta AI in the EU market, which had been delayed compared to other regions partly due to regulatory considerations.

The EU’s AI Continent Action Plan

While Meta navigates its AI data strategy, the European Commission has introduced the AI Continent Action Plan to strengthen Europe’s position in the global AI race. Launched on April 9, 2025, this initiative aims to transform Europe’s traditional industries and talent pool into “powerful engines of AI innovation and acceleration,” as outlined by President von der Leyen at the AI Action Summit in February 2025.

The plan is structured around five strategic pillars:

1. Building AI Infrastructure: The Commission plans to establish at least 13 AI factories across Europe, leveraging its supercomputing network, and create up to five AI gigafactories—large-scale facilities with massive computing power and data centers capable of training complex AI models. Through the InvestAI initiative, the EU aims to mobilize €20 billion for these gigafactories.
2. Increasing Data Access: The upcoming Data Union Strategy, expected in late 2025, aims to foster a true internal market for data, enabling scaled AI development across the EU. This includes implementing data labs within AI factories to gather and organize high-quality data from diverse sources.
3. Developing Algorithms and Fostering AI Adoption: The plan promotes AI adoption in strategic sectors including advanced manufacturing, aerospace, security & defense, energy, mobility, pharmaceuticals, biotechnology, and electronic communications. The upcoming Apply AI Strategy will target specifically increasing the currently low 13.5% adoption rate of AI in EU companies.
4. Strengthening AI Skills and Talent: To enlarge the EU’s pool of AI expertise, the Commission will focus on training the next generation of AI specialists through increased European Bachelor’s, Master’s, and PhD programs, while also working to attract and retain international and European AI talent. The AI Skills Academy, opened for proposals on April 15, 2025, will offer specialized education and training in AI and generative AI.
5. Regulatory Simplification: The final pillar addresses regulatory streamlining, particularly regarding the EU AI Act. The Commission is launching the AI Act Service Desk as an information hub offering tailored guidance, which will be especially beneficial to smaller providers. The Commission also plans to identify further measures to facilitate smooth application of the AI Act based on feedback from the Apply AI Strategy public consultation.

The AI Literacy Repository Initiative

In parallel with these developments, the EU AI Office has established a living repository of AI literacy practices following the application of Article 4 of the AI Act. This repository aims to provide examples of ongoing AI literacy initiatives to encourage learning and exchange among providers and deployers of AI systems. The practices are organized alphabetically according to their implementation status: fully implemented, partially rolled-out, or planned.

The EU AI Office has recently launched a new survey open to all organizations willing to share their experiences with AI literacy, particularly in the context of Article 4 compliance. The Office will regularly verify submissions against criteria for transparency and reliability before adding them to the public repository.

While the repository provides valuable insights into how organizations are approaching AI literacy requirements, the Commission explicitly states that replicating these practices does not automatically ensure compliance with Article 4. The repository covers various sectors and business sizes, from micro (1-15 employees) to large (250+ employees).

Most organizations in the repository report using training or e-learning programs for staff, often segmented into different learning levels or topics such as AI fundamentals, advanced AI concepts, AI regulation, or AI in specific business contexts. Some have developed separate programs for technical and non-technical staff, with approaches varying between voluntary and mandatory participation.

Why It Matters

For Meta and Technology Companies: The resumption of AI training using European user data represents a critical step in Meta’s ability to compete in the EU market. Without access to region-specific data, AI models can lack the cultural context and linguistic nuance needed to serve European users effectively. This development may set precedents for how other technology companies navigate EU data regulations while developing competitive AI solutions.

For European Businesses: The AI Continent Action Plan signals Europe’s commitment to close the AI gap with competitors like the US and China. The comprehensive approach—spanning infrastructure, data, skills, and regulation—aims to create an environment where European businesses can develop and deploy AI solutions that reflect European values while remaining globally competitive.

For EU Policymakers: These initiatives highlight the delicate balance EU regulators are attempting to strike: maintaining Europe’s leadership in digital rights and privacy protections while creating conditions that foster innovation. The AI Act Service Desk and the simplification measures indicate recognition that compliance support is essential for the success of EU’s regulatory framework.

For Global AI Development: Europe’s approach represents a distinct “third way” in global AI governance—neither the more market-driven approach of the US nor the state-directed model of China. The success or failure of initiatives like the AI Continent Action Plan will influence how other regions approach the balance between innovation and regulation in AI development.

While Europe’s investments in AI infrastructure remain modest compared to those in the US and China, these coordinated efforts to build capacity, improve regulatory frameworks, and support implementation could help European organizations maximize the impact of available resources. The coming months will be critical in determining whether these initiatives successfully accelerate AI adoption across the European economy.