Updated commercial terms taking effect October 30 appear to let Meta crawl publisher websites and use that content to train its AI

Meta has rewritten the fine print for businesses on its platforms. According to Nieman Lab, the new language appears to let the company crawl and scrape websites “associated” with those businesses, and that includes online publishers.

The updated terms take effect on October 30, 2026. Once they do, Meta says it can use that outside content to “develop, improve, research,” and “train” its AI products. For newsrooms that have spent recent years negotiating payment for their archives, that’s an uncomfortable sentence.

The changes sit inside Meta’s commercial terms of service. That is the rulebook for business accounts, not for the average person posting vacation photos.

The scope is broad. It covers any business account that manages pages or groups, runs ads, or uses Meta’s business tools across Facebook, Instagram, WhatsApp, and Messenger.

The terms permit Meta to use automated tools, such as web scrapers and crawlers, to collect content tied to those accounts.

The stated uses are wide-ranging. They include providing and developing Meta’s products, and training them. The terms define those products to include “AI at Meta,” so the AI angle is written directly into the document.

Nieman Lab points out what this could mean in practice. The language seems to give Meta a way to train its models on publisher content without first signing a licensing agreement.

Most major news publishers keep business accounts with Meta. They use them to promote stories, manage pages, and run ads. Under the reading that has emerged, that ordinary marketing presence may be enough to put a publisher’s own website within Meta’s reach.

Daily. Free. Join 34,000+ readers across crypto, finance, and policy.

The part drawing the most attention is the missing off switch. The new terms include no opt-out.

Businesses that keep using Meta’s services after the effective date will be treated as having accepted the terms. Avoiding them means fully ending business use of Meta’s platforms.

Meta has notified business users of the changes through its platforms. That outreach has already raised questions about how enforceable these provisions will be, and how far an agreement about using an ad tool can extend over a separate website.

Publishers have pushed hard for payment. In the emerging model, AI developers license news content directly, which gives media companies a new revenue stream as advertising and traffic from social platforms get less dependable.

Meta’s new language appears to sidestep that model. Rather than negotiating with each publisher, the company would rely on terms that publishers effectively accept just by staying active on its platforms.

That shift has prompted legal scrutiny. The concerns include possible antitrust issues and the gradual erosion of the licensing market that content creators have been trying to build.

For publishers, the stakes are mostly economic. Licensing deals depend on scarcity. If a dominant platform can get the same content through a contract clause, the bargaining position of every other publisher weakens.

There is also a competitive concern. AI products trained on news reporting can answer questions that readers once clicked through to a publisher’s site to find. A publisher could end up helping train a tool that competes with it for the same audience.

Several questions will decide how this unfolds. The first is enforceability: whether accepting a business tool agreement can lawfully authorize scraping a separate website is the kind of issue that tends to end up in court.

The October 30 date gives publishers a narrow window to decide where they stand. After that, staying put is treated as agreement.