Meta's updated commercial terms take effect on October 30, 2026. According to Ars Technica's review published on October 16, the terms apply to all business accounts and reach beyond what companies post on Meta's platforms to the websites linked to those accounts. There is no opt-out: a business that disagrees is pointed toward stopping its use of Meta's services.
The dispute centres on scraping permission. The new terms authorise Meta and its agents to use automated software such as crawlers to access, index, cache, analyse and collect content from associated domains, pages and URLs, pulling updates on a continuing basis, for purposes listed as providing, developing, improving, researching, training and maintaining Meta's products. The text never says AI outright, but the word training is what alarms legal experts. Three of them told Nieman Lab that it is unclear whether courts would enforce such a grant, yet Meta might later rely on these terms to claim a right to scrape news sites for AI training, avoiding content licensing fees that could run to many millions of dollars.
Two words stand out in the new grant: test credentials
Ars Technica compared Meta's current documents and found that permission to create test credentials does not appear in the current Commercial Terms, Product Catalog Terms or Ad Creative Generative AI Terms. Similar wording surfaced in this month's Self-Service Ads Terms, and the Seller Agreement has carried a related passage, but only for e-commerce sellers optimising product listings. Experts worry test credentials could be used to get past paywalls automatically. Seen another way, narrow permissions that used to sit scattered across documents, each tied to a single purpose, have been consolidated into one broad grant. That is also Meta's own account: a spokesperson said the company's crawler use has not changed, that the language previously appeared across several documents including the generative AI ads terms and is now simply merged into one place, and the company blog says the change reduces overlapping agreements and improves clarity. The spokesperson also told Nieman Lab that Meta has no current plan to use the new terms as a legal basis for ignoring robots.txt, and that site owners can still manage access through robots.txt.
Publishers are hit hardest, because nearly all of them hold a business account
Reporting on the change notes that almost every news organisation holds at least one Meta business account, which means they are covered by default. Cornell law professor James Grimmelmann warns that Meta could use the scraped content on its own platform in ways that compete with publishers, and other experts compare the arrangement to tying: use the platform's business services, and hand over scraping rights to your website content in the same deal. With Meta also recently starting to charge users for posting links, publishers are squeezed from both sides, paying for traffic while potentially surrendering content through the terms. Stephen Beale, founder and editor of The Steampunk Explorer, which depends on Facebook traffic, is already considering leaving the platform.
Before October 30, businesses have three checks to run
First, inventory which Meta business accounts the company holds and which domains they are linked to, so the actual coverage is known. Second, review those sites' robots.txt settings and paywall mechanics, and be precise about what the company wants to block and what is actually blocked today. Third, keep dated records of those settings: if scraping later appears to go beyond what robots.txt allows, that record is the basis for any dispute. Whether the terms hold up in court, and whether publishers push back collectively, will only become clear after they take effect, but the window for self-checking is just the days that remain before October 30.
