Meta's October 30 Terms Remove 'Train' but Still Let It Scrape Business Sites Broadly
Meta published revised commercial terms Friday afternoon. They take effect October 30. The headline change removed the word “train” from the list of permitted activities. What remains has legal experts concerned.
What Changed
The prior terms allowed Meta to use business content for all Meta products. The revised version limits those uses to “advertising and business products.” The word “train” is gone from the enumerated permissions.
What did not change: the terms still authorize Meta to scrape websites broadly. The specific concern is that any business operating a website and using Meta’s commercial tools — ad pixels, business manager, Shops integrations — has agreed to terms that, per the revised language, let Meta access and process content from that site for advertising and business purposes.
Legal experts reviewing the language say this effectively supersedes robots.txt for participating sites. A robots.txt directive tells crawlers not to come. A business agreement may override that at the contract level.
The Publisher Problem
News publishers are the most exposed. Many run Meta pixel or use Meta’s business tools for audience development and advertising. Under the revised terms, those tools carry a data-access clause that legal reviewers say is broad enough to cover content scraping for ad targeting and product development.
The removal of “train” narrows the explicit AI training use case. It does not foreclose using scraped content to train ranking, recommendation, or ad prediction systems that Meta frames as “business products” rather than foundation models.
Legal experts, including Cornell law professor James Grimmelmann, argued that agreed-upon commercial terms create a contractual license Meta could use to override robots.txt directives for participating sites.
Stakes
The practical scope depends on how Meta enforces the terms and what courts eventually decide about contract-based scraping rights. But the directional pressure is clear: by tying data access rights to business tool agreements rather than standalone crawl policies, Meta can bypass content publisher opt-outs that apply to anonymous crawlers.
Publishers who do not want content used under these terms face a binary: remove Meta’s business tools from their sites, or accept the terms. That is not a comfortable trade for outlets that depend on Meta’s ad infrastructure for revenue.
Meta’s Pattern
This is consistent with how Meta has expanded data access rights incrementally. The Muse agent was blocked by Amazon for unauthorized retail access in September. Now the commercial terms revision shifts the legal framing — from unauthorized access to permissioned scraping embedded in business tool agreements.
The October 30 effective date gives publishers roughly two weeks to respond.