Meta Removes AI Training Language, but Crawling Dispute Remains

Meta removed language from forthcoming commercial terms that said content from websites connected to business accounts could train Meta products. The revision followed objections that the wording could authorize broad scraping of news and business sites.

The terms take effect October 30. Reporting on the revised text says Meta removed the word "train" from the disputed uses and narrowed the purpose to advertising and business products. Meta said the update was intended to improve product information used in ads and recommendations, not to scrape news sites for AI training.

The change reduces the immediate scope, but does not settle every issue. Meta may still crawl sites linked to business accounts for commercial functions, and publishers may dispute whether contractual permission overrides technical signals such as robots.txt. No court ruled on the abandoned wording, so the revision is not a legal finding.

For Meta shareholders, training-data access carries licensing, regulatory and reputation costs. Removing explicit training language may lower backlash but could constrain a low-cost data route. The next evidence is the final October 30 text, enforcement practice and any publisher action.