Meta has deleted the word "train" from its business terms. The revision lands Oct. 30, and it follows a backlash from publishers who objected to the company's data-scraping practices.

That's the decision. Now the mechanism — and why legal experts aren't reassured.

What did Meta actually change?

The company stripped references to training from the terms that govern its business accounts. Publishers had pushed back hard after discovering those terms appeared to license Meta to scrape their sites — pulling content and data for use in its AI systems.

So Meta blinked. Sort of.

The removal of "train" reads like a concession. It isn't, necessarily. Legal experts who reviewed the revised language say the new terms, effective Oct. 30, could still allow Meta to scrape sites connected to business accounts. The word is gone. The exposure isn't.

Why does one word matter so much?

Business terms are contracts. When a publisher or any company holds a business account on Meta's platforms, those terms govern the relationship — including what Meta may do with data it collects from that company's own websites.

"Train" was the flashpoint. Its presence suggested Meta could use scraped publisher content to train its AI models without separate negotiation or payment. Publishers, already fighting synthetic content built from their archives, objected publicly.

Removing the term narrows the stated purpose. But the experts' reading is blunt: the revised wording leaves room for scraping anyway. The contractual door Meta closed with one hand, it propped open with the other.

What does the Oct. 30 date mean for publishers?

The revised terms take effect Oct. 30. Until then, the current terms govern. After that date, business account holders operate under the new language — the version without "train," but with wording that, per legal experts, could still permit scraping of sites tied to those accounts.

For publishers, the practical takeaway is narrow but real:

• The explicit reference to AI training is out of the business terms.
• Scraping exposure remains, in the experts' assessment.
• Any site linked to a Meta business account stays within reach of the revised language.

What's the strategic logic here?

Meta needs data. Its AI ambitions run on it, and the open web — including publisher sites — is the richest available source. Publishers, meanwhile, want compensation and control. Regulators in multiple jurisdictions are circling the question of who owns training data.

Against that backdrop, a quiet terms edit does three things at once. It defuses the immediate publisher backlash. It avoids conceding any legal position. And it keeps the scraping pathway technically intact, according to the lawyers who parsed the new text.

That's a negotiation move, not a retreat.

What should companies with Meta business accounts do?

The obvious first step: read the revised terms before Oct. 30. Anyone operating a business account — publisher or not — will be bound by them.

The second step is harder. If the experts are right that scraping exposure survives the edit, then companies concerned about their site data need to weigh the value of the business account against the reach of the terms attached to it. Disconnecting a site from a business account is, on the experts' reading, the clearest way out of the revised language.

Meta hasn't surrendered the scraping question. It has restated it — in smaller type, so to speak. The Oct. 30 deadline gives publishers and business account holders a fixed window to decide how they respond.