Meta moves its website scraping terms to 20 November

In October 2026, Meta changed the terms for its business accounts. The terms let Meta crawl linked websites, and publishers think that Meta can use their content to train AI.

Earlier: Meta terms let it crawl business websites to train AI

Meta moved the start date of its new business terms from 30 October to 20 November this week. The terms let Meta use automated tools to crawl websites that are linked to a business account. Publisher groups on each side of the Atlantic told Meta to stop, and made complaints to regulators. The News Media Association in the UK tells that the move is only a stop and not a solution.

What changed

After one day, Meta removed the word train from the terms. It kept the content that it uses in its advertising and business products. A Meta spokesperson tells that the terms are not about scraping news sites.

The proof

The source gives no test results. Three legal experts that Nieman Lab spoke to told publishers of the risk. Legal experts tell that the new text continues to let Meta do mass scraping.

What is not known

The source does not tell if Meta will change the text again before 20 November. It does not name the regulators that have the complaints.

Insight: The business accounts connect the site access and the agent rules

Meta tells that its terms apply to business accounts. In October 2026, Meta and Sierra started work on an open standard for personal AI agents. Shopify, Stripe and Walmart are partners. This reads as a control point. Meta holds the business accounts, and the same companies give site access and use its agent rules. Monitor if the text changes before 20 November.