Meta's New Terms Let Its AI Scrapers Crawl Publishers' Sites

Meta will gain the right to crawl, index and collect content from websites "associated with" any Meta business account when its new commercial terms take effect on Oct. 30, and the company will offer no opt-out. Legal experts cited by Nieman Lab say the change turns every publisher using Meta Business Suite into a silent supplier of training data. Businesses that refuse face removal from the platform.

The stakes are considerable. AI labs need high-quality text to train large language models (systems that generate and summarize text by predicting word sequences from vast datasets), and journalism — fact-checked, edited, structured — ranks among the best sources available. Personal blogs and open-web text do not match it.

What exactly do the new terms authorize?

The terms apply to any account using the Meta Business Suite dashboard, through which companies manage Instagram profiles, Facebook pages, Messenger, and WhatsApp channels. The contractual language is broad. It reads:

"You authorize us (or any entity acting on our behalf) to use automated software (such as web scrapers, bots, robots, spiders, crawlers, and user agents) and create test credentials to access, index, cache, analyze, collect, retrieve, or crawl the domains, web pages, or URLs (and any content available through them) that are associated with your products or services... and to periodically pull updates from such domains, web pages, or URLs."

The clause does not stop at collection. It continues: "Meta may use any content from such domains, web pages, or URLs to provide, develop, improve, research, train, and maintain the integrity of, the Meta Company Products, services, and features."

Three elements stand out:

• Automated access: scrapers, bots, spiders, crawlers, and user agents are all named explicitly.
• Test credentials: Meta may create its own login credentials to reach content.
• AI training: the word "train" appears in the list of permitted uses, directly.

The language covers not only domains publishers list in their accounts but also URLs surfaced "through your or users' use of Meta Company Products and services."

Why is Meta doing this now?

Nieman Lab reports that Meta could be using the scraping language to avoid entering licensing deals with websites and news publications. The contrast with competitors is sharp. Several AI labs have already signed licensing agreements with news outlets to secure training data legally.

Meta's stated position is narrower. According to the Verge, the company says it is merely "consolidating" its commercial and self-serve ad terms — folding pre-existing conditions into a single document. In Meta's framing, nothing new is being claimed. Legal experts quoted in the Nieman Lab report disagree, reading the terms as newly authorizing web-crawling on publication sites.

Meta did not immediately respond to a request for comment.

What does this mean for publishers?

Publishers face a binary choice on Oct. 30: accept the scraping terms or lose access to Meta Business Suite and, by extension, the platform's distribution. There is no middle path and no opt-out mechanism.

That choice lands amid an already hostile environment for news organizations. AI-powered search features have cut traffic to news sites, and with it advertising revenue, which is why publishers — including Ziff Davis, ZDNET's parent company — have filed copyright lawsuits against AI companies. The suits target both model training and the delivery of answers to search queries without attribution.

For a publisher, the calculus is unenviable. Facebook and Instagram remain major distribution channels, and abandoning them carries a real audience cost. Staying means granting a blanket license to automated collection.

How does this fit the wider data fight?

The terms change is the latest move in a contest over who controls the raw material of AI. Licensing deals compensate publishers for training data. Scraping, authorized through terms of service, costs the AI developer nothing per document and leaves publishers without leverage once they have clicked agree.

The journalistic-content problem is specific. Editors and fact-checkers make news archives valuable in a way that arbitrary web text is not, and publishers who have spent decades building that value now watch it become an input to models built by the platforms they depend on for distribution.

Unless Meta offers an opt-out, reverses the clause, or negotiates licensing arrangements before Oct. 30, every business on Meta Business Suite will effectively make that decision by default — and the data flowing to Meta's crawlers will begin with the first scheduled post after the deadline.

By Marcus Bennett