Meta edits business terms, but experts still worry about AI scraping

In short: Meta changed its business terms to remove a line about training AI on website content, but legal experts say the updated wording may still allow wide scraping of news sites.

What happened

Meta recently updated its terms for business accounts, which are rules that apply when a company uses Meta services. An earlier version said Meta “may use any content” from websites linked to a business account to help “develop, research” and “train” Meta products. The terms are set to take effect on October 30.

After strong criticism from publishers, Meta removed the specific word “train” from that section. Some publishers feared the terms could be used to justify taking articles and other site content for AI systems without paying.

However, legal experts quoted by Nieman Lab say the changes may not address the core issue. Cornell law professor James Grimmelmann pointed to language that still covers websites “associated with your products or services,” which he says could be read broadly enough to include whole sites. Attorney Matt Topic also argued that letting Meta use scraped content to “develop” a “business product” could still include AI training, since “develop” is a broad word and “business product” is not clearly defined.

Meta told The Verge that the update is not meant to scrape news content for free. Meta also said content owners can use robots.txt, a small file on a website that acts like a “do not enter” sign for automated scrapers, to signal their preferences.

Why it matters

This dispute affects how much control publishers and other website owners have over their work online. If terms like these are interpreted broadly, a company could argue it has permission to collect large amounts of content, even when a site tries to block it. That could impact where AI tools get their information, and whether creators get a say or compensation.

Source: The Verge AI