Meta's Updated Terms Could Enable AI Training on Publisher Content
Meta Platforms has quietly updated its commercial terms in a way that could have significant implications for online publishers. According to an analysis by Nieman Lab, the new terms—set to take effect on October 30th—appear to grant Meta permission to crawl and scrape websites "associated" with businesses that use its platforms.
The updated language explicitly states that Meta can use this scraped content to "develop, improve, research," and "train" its AI products. While a Meta spokesperson described the changes as merely a "consolidation" of existing commercial and self-serve advertising terms, legal experts interviewed by Nieman Lab have raised concerns about the broader scope of the new provisions.
The most contentious aspect involves how these commercial terms might interact with existing website blocking mechanisms. Industry-standard tools like robots.txt files have traditionally allowed publishers to indicate which automated systems can and cannot access their content. However, legal experts suggest that Meta's updated commercial terms could be interpreted as superseding such instructions, creating a potential conflict between contractual agreements and technical protocols.
For news publishers already grappling with questions about how their content is used to train AI systems, this development adds another layer of complexity. Publishers have increasingly sought compensation or clearer consent frameworks for AI training use cases, but Meta's new terms may sidestep some of those negotiations by relying on the commercial relationship with advertisers and platform users.
The changes come as regulatory scrutiny of AI training practices continues to intensify globally, with several jurisdictions considering or implementing frameworks that would require greater transparency and consent for using copyrighted or published content in AI development.