Meta has restricted the use of public Instagram and Facebook posts to train its AI models in the United Kingdom and parts of Europe following regulatory pressure. The company confirmed it would no longer use public content from adult users in these regions to train its "Meta AI" features, responding to interventions from the Irish Data Protection Commission (DPC).
Regulatory Intervention in the European Union
The decision follows a request from the Irish DPC, which acts as Meta’s lead privacy regulator within the European Union. In June 2024, the DPC asked Meta to pause its plans to use public data from Facebook and Instagram users to train its large language models. The regulator expressed concerns regarding the transparency and legal basis for processing this personal data under the General Data Protection Regulation (GDPR).
According to a Meta spokesperson, the company agreed to pause the training of its AI models on public user content to address the DPC’s requests and ensure compliance with European privacy standards. While Meta maintains that its approach aligns with GDPR requirements, the pause marks a significant shift in how the company develops its generative AI tools for European markets.
Impact on Meta AI Features
The restriction specifically targets the training data sets used for Meta AI, the company’s suite of generative AI tools integrated into its social media platforms. By removing public posts from European users from the training pipeline, Meta must now rely on alternative data sources to refine its models in those jurisdictions.
This change does not apply to private messages or the content of users under the age of 18, which Meta has stated it does not use for AI training. However, the policy change creates a distinct regional difference in the development of Meta’s AI, as the company continues to utilize public posts for training in the United States, where there is no equivalent federal mandate halting the practice.
Data Privacy and Generative AI
The conflict highlights the ongoing tension between AI developers and data protection authorities over the use of "publicly available" social media data. While Meta argues that public posts are a standard part of training datasets for large language models, privacy advocates and regulators argue that users did not explicitly consent to their personal data being repurposed for AI training when they created their accounts.
The Irish DPC’s intervention is part of a broader trend of regulatory scrutiny regarding how technology firms handle personal data in the age of generative AI. As of mid-2024, the company has not provided a definitive timeline for when—or if—it will resume using European public data for AI training, stating only that it is working with the DPC to address the outstanding concerns.
Summary of Regional Training Policies
| Region | Status of Public Post Usage for AI Training |
|---|---|
| United States | Active |
| United Kingdom | Paused |
| European Union | Paused |
The pause in Europe serves as a precedent for how global technology companies may be forced to segment their AI development processes to accommodate varying regional privacy laws. For users in affected regions, the change represents a victory for opt-out rights, though it also underscores the complexity of regulating data usage in rapidly evolving AI systems.