Meta revises business terms, but legal experts still see risk of AI training on news sites

Meta, the company behind Facebook and Instagram, has reworked its new terms for business accounts to remove wording that appeared to let it freely train artificial intelligence (AI) models on news sites, including The Verge. Legal experts, however, say the changes do not resolve their core concerns. The terms take effect on October 30th.

What the original terms said

The first version stated that Meta could use any content from websites associated with a business account to "develop, research" and "train" its products. Experts told Nieman Lab this could let the company argue that its commercial terms override other blocking instructions, such as the robots.txt file, which has long told scrapers, including search engines, what they may crawl.

Changes and criticism

After strong backlash from the publishing industry, Meta removed the language about "training" products with scraped content. Yet James Grimmelmann, a professor of digital and information law at Cornell Law School, said the change does nothing to limit the phrase "associated with your products or services," which would allow entire sites to be scraped.

Matt Topic, a media and intellectual property attorney at Loevy & Loevy, said the revised text still permits the use of scraped content to develop any "business product." He noted that the term does not appear to be defined, that Meta is presumably building AI business products, and that "develop" seems broad enough to cover "train."

Meta's position

The company maintains that the terms are not meant to scrape news content for free. Meta spokesperson Daniel Roberts told The Verge that content owners can use robots.txt to signal their preferences, and that the update is not about scraping news sites and does not give Meta new rights that supersede those preferences.