Facebook owner Meta seeks to train AI model on European data as it faces challenges with EU privacy laws
Meta, the parent company of social media platforms like Facebook, Instagram, and WhatsApp, is looking to utilize data from users in Europe to enhance its artificial intelligence models. This move comes as Meta strives to compete with tech giants such as OpenAI and Google, while grappling with concerns regarding data privacy.
The company highlighted the importance of incorporating the diverse languages, geography, and cultural nuances of European users into its AI models. To achieve this, Meta aims to use publicly available data from European users to train its Llama AI large language model.
Challenges with EU Data Privacy Laws
Meta's AI training endeavors are being impeded by the strict data privacy regulations imposed by the European Union. These laws empower individuals to have control over the usage of their personal information, posing a challenge for Meta's data training activities.
Vienna-based group NOYB, spearheaded by activist Max Schrems, recently lodged complaints with 11 national privacy authorities regarding Meta's AI training plans. The group urged regulators to intervene and halt Meta's efforts to train the next generation of Llama's AI model.
Training AI Language Models
AI language models like Llama rely on vast datasets to enhance their predictive capabilities. By analyzing extensive data, these models can anticipate the most likely succeeding word in a sentence. With each iteration, newer versions of AI models become more intelligent and proficient.
While Meta has integrated its AI assistant feature into platforms like Facebook, Instagram, and WhatsApp for users in various countries, including the U.S., it has notably excluded European users from this feature.
Ensuring Cultural Relevance
Stefano Fratta, Meta's global engagement director for privacy policy, emphasized the significance of training AI models on public content shared by Europeans on social media platforms. Fratta stated that neglecting to train models on relevant European data could result in AI features that lack a nuanced understanding of regional languages, cultures, and trending topics.
Fratta clarified that Meta will not utilize private messages or content from European users under the age of 18 for training purposes. The company has proactively informed European users about its plans and provided an opt-out option through notifications and emails.
Moving Forward
Meta's updated privacy policy, set to come into effect on June 26, indicates the imminent commencement of training for the next AI model. While Meta faces challenges with EU data privacy laws, it is striving to ensure that its AI models are well-informed by Europe's cultural diversity and historical contributions.
Other tech companies like Google and OpenAI have already leveraged European data for training their AI models, highlighting the importance of cultural relevance in AI development.
Technology / Internet / Artificial Intelligence / Litigation and Regulation / European Union
