Meta has revised its forthcoming commercial terms after language about using website content raised concerns that publishers were granting permission for AI training. The Verge reported the removal on October 17; TechCrunch separately covered the change. Meta’s published preview lists October 30, 2026 as the effective date.

The disputed sentence said Meta could use content from associated domains, pages or URLs to “provide, develop, improve, research, train, and maintain the integrity of” its products and services. Legal experts cited in the reporting interpreted that wording as potentially authorizing broad use of publishers’ websites, including for model training.

What the revised terms say

The current preview still authorizes automated access, indexing, caching and crawling of covered websites. However, the crawling provision now specifies developing and maintaining advertising and business products, alongside protecting the integrity and security of Meta products. The separate sentence containing the explicit reference to training no longer appears.

That is a narrower contractual description, not a withdrawal of all permission to crawl websites. The commercial terms cover activities including advertising, managing Pages and using business tools.

Meta denies an operational change

In a statement quoted by The Verge, spokesperson Daniel Roberts said the consolidated crawling terms applied only to advertising and business products. He said website owners could continue signaling preferences through robots.txt and that the update did not grant new rights overriding those preferences.

For publishers and businesses, the verified development is a revision to pending contract language. It does not establish that Meta has stopped collecting web content or changed how it trains AI models.