Meta’s New Terms Raise Questions Over Free AI Training on Publisher Sites

Meta’s revised commercial terms have triggered a dispute over how business relationships with the platform may affect the use of website content. The controversy is especially relevant to publishers whose sites are connected to business accounts on Facebook, Instagram, WhatsApp, or Messenger.

What the terms say

The terms authorize Meta, or entities acting on its behalf, to use automated software such as scrapers, bots, spiders, crawlers, and user agents. These tools may access, index, cache, analyze, collect, retrieve, or crawl domains, pages, and URLs associated with a business’s products or services, or made available through the use of Meta’s products. The language also permits Meta to pull updates from those locations periodically.

The most consequential wording concerns the possible uses of the material. Meta says content from those domains may be used to provide, develop, improve, research, train, and maintain its products, services, and features. Read broadly, that could cover much more than retrieving product details for advertisements. It could also be interpreted as permission to use material published on a company’s website for AI-related development.

Why publishers are concerned

News organizations generally distinguish between ordinary web access, licensed syndication, and the use of articles as training data for generative AI. The new language appears to connect those issues to a business’s continued use of Meta’s commercial products. Publishers may worry that maintaining a business presence on Meta could be treated as acceptance of a broad content license.

Another question involves robots.txt, a standard mechanism websites use to communicate preferences to crawlers. Legal experts cited by Nieman Lab suggested that Meta could potentially argue that its commercial terms supersede certain blocking instructions. Cornell digital and information law professor James Grimmelmann said the language could give Meta permission to train AI on content from publisher sites and possibly deploy that material in ways that compete with those publishers.

Meta disputes that interpretation. Spokesperson Daniel Roberts said the company was consolidating its commercial and self-serve advertising terms, not changing how its web crawlers operate. He also said content owners can still use robots.txt to indicate their preferences. Meta communications executive Andy Stone argued that the update is intended to let the company retrieve accurate product information from third-party websites for on-platform advertising and content recommendations, rather than target news sites.

A later clarification

The terms reportedly take effect on October 30. The material also notes that Meta later updated the wording and said the consolidated crawling provisions applied only to advertising and business products, adding that this limitation had been clarified in the terms themselves. That clarification narrows the apparent scope, but it does not settle every question about what counts as associated content or how business terms interact with a site’s technical instructions.

Meta has already reached AI licensing agreements with CNN, Fox News, USA Today, and People Inc. A broader contractual authorization could give it access to a wider pool of material without negotiating a separate deal with every publisher. At the same time, businesses that reject the new policies may lose access to Meta products for business or commercial purposes, according to the terms.

Why it matters

The dispute illustrates how AI data governance is moving beyond dedicated licensing contracts. Advertising agreements, platform access, crawler policies, and publisher rights are becoming part of the same negotiation. Publishers may need to review how their sites are linked to business accounts, how crawler preferences are expressed, and whether their platform contracts contain language that could be read as consent to AI training.

For Meta, the central challenge is to draw a clear line between collecting product information for advertising, recommending content, and training AI systems. Until that boundary is made explicit, broad contractual language is likely to keep raising concerns among publishers about free access to their reporting and the possibility of platform-generated products competing with the original sources.

Source: The Verge AI