Meta Removes “Train” From October 30 Terms, but AI Scraping Fears Persist
Meta removed the word “train” from its October 30 business terms, but legal experts and publisher groups say broad scraping rights and robots.txt gaps remain.
Summary
On October 20, 2026, Meta removed “train” and “any content” from commercial terms taking effect October 30, without notifying users of the revision. The original language covered website content linked to business pages and permitted its use to provide, develop, improve, research, train and maintain any Meta product. Meta may now “periodically pull updates” to provide, develop, improve and research advertising and business products, maintain security, support ads, display product information and connect businesses with interested users.
Meta says the consolidated terms apply only to advertising and business products, do not change its crawlers, target news sites or override robots.txt preferences. Matt Topic of Loevy & Loevy says the undefined phrase “business product” could include AI and that “develop” may encompass training. Cornell Law School’s James Grimmelmann says language covering content associated with products or services could still expose entire sites. University of Wisconsin professor Benjamin Sobel warns robots.txt remains voluntary without an enforceable commitment. Experts say Meta could avoid millions in publisher licensing fees while conditioning audience access on scraping consent. Theo Bamber of the UK News Media Association, Angela Mills Wad of the European Publishers Council and Danielle Coffey of the News/Media Alliance, the largest US news publishing trade association, want regulators to block the terms over consent, compensation and market power.
Positives
- Meta removed “train” and “any content” from the commercial terms before their October 30 implementation.
- Updated language limits the stated purposes to advertising and business products, product security and services benefiting business users.
- Meta says the revision creates no new crawler rights and does not override content owners’ robots.txt preferences.
Risks & concerns
- Meta did not notify business users after revising terms scheduled to take effect October 30.
- Matt Topic says the undefined phrase “business product” may cover AI, while “develop” could still permit model training.
- James Grimmelmann says content associated with products or services remains broad enough to expose entire websites.
- Benjamin Sobel says robots.txt instructions can still be ignored because compliance is voluntary without an enforceable commitment.
- Experts warn Meta could avoid millions in licensing fees by requiring businesses to accept scraping for access to its audience.
- Publisher groups in the UK, European Union and US are urging regulators to prevent the terms from taking effect.