Meta’s AI could still scrape news sites, legal experts say

Meta changes its terms, but worries remain

Meta updated the terms for its business accounts. The first version seemed to let Meta train its AI on news sites, such as The Verge. After strong pushback from publishers, Meta removed the word "train." Legal experts say the new version may still allow the same thing.

Key facts

• The new terms go into effect on October 30.
• The first version said Meta "may use any content" from sites linked to a business account.
• The old wording was "develop, research" and "train." Meta removed "train."
• Meta says content owners can use robots.txt to say what they allow.

What is robots.txt?

Robots.txt is a small file that a website uses to tell automatic programs, called scrapers, what they may copy. Search engines like Google follow it too. Scraping means copying content from a site automatically. Experts worry that Meta's terms could let it ignore these rules.

Why it matters

News sites make their money from readers and ads, and their stories are used to train AI. If AI companies can copy news without clear permission, publishers may lose money and control. The debate shows how unclear the rules for AI and online content still are.

What to watch

• Whether Meta makes more changes to the wording of its terms.
• Whether legal experts or publishers take formal action.
• How the October 30 start date affects news sites and their robots.txt rules.