Meta pushes back on new AI data terms for business content as UK and France scrutiny intensifies

Meta Platforms has delayed new terms governing how it may use data from business users for artificial intelligence model training, a move that reflects mounting pressure from regulators and industry groups in Europe. The company was preparing to introduce terms that would have allowed broader extraction of business content for training, but the rollout has been pushed back, according to a report carried by Yahoo Finance.

The reported change goes to the heart of how companies build AI systems at scale. Instead of relying only on public data or user-provided training datasets, platforms can also use content they host, under specific usage rights. Meta’s proposed terms would have expanded the company’s ability to use content connected to its business ecosystem for training purposes.

European publishers and related industry groups have argued that such training rights should be constrained, warning that scraping and reuse of publisher-adjacent content can undermine consent and copyright expectations. In the UK and France, the report points to pressure building around the legality and fairness of how large platforms acquire and repurpose data for AI systems.

In response to that environment, Meta’s decision to delay suggests it is seeking additional time to address objections, negotiate with stakeholders, or align the rollout with evolving legal interpretations. While the company has not publicly outlined, in the Yahoo report, detailed conditions for when the delayed terms will take effect or what specific edits might follow, the delay itself indicates sensitivity to the compliance risk and reputational friction tied to AI training practices.

The dispute also fits a broader European pattern in which authorities and industry players have challenged how tech companies train AI models. Across multiple countries, governments and publisher groups have increasingly emphasized transparency, consent, and intellectual property protections, especially when content is generated by third parties but processed or analyzed by platforms.

Meta, for its part, has already been positioning its services around AI capabilities, including tools that rely on large-scale data processing. But AI ambitions face practical limits when the legal foundation for data use is contested. A delay in AI data terms typically shifts timing risk, not only for policy changes but also for internal engineering plans that depend on clarified permission to use certain datasets.

What Meta did not disclose in the Yahoo report is as important as what it did. The reported information does not specify the new expected timeline for the revised AI terms, whether the company will narrow the scope of business-content training rights, or how it will handle opt-outs, consent mechanisms, or licensing questions raised by publishers and industry groups. Without those details, it is unclear whether the delay is temporary administrative postponement or a substantive rework of the underlying data-use approach.

Looking ahead, the key question for markets and for the affected industries is how Meta will resolve the policy and legal tension that prompted the delay. Observers will likely watch for any follow-up announcement describing revised AI data terms, any response from European publisher groups, and any regulatory guidance that clarifies whether and how business-content reuse for model training is permitted. For now, the latest announcement is that Meta is not moving forward on the earlier schedule.