Meta Drops the Word 'Train' From Commercial Terms Ahead of October 30

Meta has deleted the word "train" from its new commercial terms of service, days before the terms take effect on October 30. The company did not announce the change. A Meta spokesperson confirmed that users received no alert about the revised language.

The edit follows a week of backlash. Legal experts warned that Meta's "streamlined" commercial terms appeared to grant the company sweeping permission to scrape websites linked from business pages. Talking to Nieman Lab, experts suggested Meta could use the policy to avoid millions in licensing fees from news organizations, by requiring business users to hand over AI training data simply for operating a page.

Meta's spokesperson provided a statement to Ars Technica. The statement acknowledged Nieman Lab's initial report. It did not address the follow-up report, which reiterated the experts' concerns. That silence leaves the central question unanswered.

What does the word change actually change?

Removing "train" does not remove the underlying rights grant. The source material gives no evidence that Meta narrowed any clause beyond the wording. The company has published no clarification of what the remaining language permits.

This matters for chip and data-center watchers. Meta's AI infrastructure buildout consumes training data at industrial scale. Any terms that lower the cost of acquiring that data — by making it a condition of running a business page — change the economics of model development directly.

The reported stakes are financial. Experts told Nieman Lab the terms could help Meta sidestep millions in licensing fees owed to news organizations. Those figures are expert estimates, not audited numbers. No party has disclosed what Meta currently pays news publishers under existing deals.

Why the silent edit matters

Meta confirmed it did not alert users to the changed language. That choice comes at the worst possible moment for the companies affected. The terms activate October 30. Business owners reviewing the old wording today may not know the wording has moved.

Confusion serves no one except the party writing the contract. Regulators in multiple jurisdictions are already scrutinizing how platforms obtain AI training data, and a quiet edit to contested terms invites exactly that scrutiny. Meta's statement to Ars did not explain why the word was dropped or who decided.

The experts' core claim also remains untested. No court has ruled on whether the commercial terms would actually shield Meta from licensing obligations. Until one does, the "millions in avoided fees" figure stays an analyst claim, not a confirmed outcome.

What to watch on October 30

Three things will indicate whether this edit is substance or cosmetics:

• Whether Meta publishes a plain-language summary of what data rights the terms grant.
• Whether any news organization or publisher group challenges the terms before or after the effective date.
• Whether regulators in the EU or elsewhere open a review of the data collection the terms permit.

None of those steps depends on Meta's goodwill. Publishers have legal standing and motive, given the estimated fee exposure. The October 30 deadline gives them a fixed date to act on.

For now, the commercial picture is unchanged in one respect. Meta still holds the drafting pen, still faces no confirmed liability, and still gets its terms live on schedule. Whether the deleted word survives contact with publishers and regulators is the question the next quarter will answer.