From October 30, Meta will require business accounts to let it train AI on content from their linked websites, not just their social media posts.
The update has no opt-out. Instead, the terms suggest that businesses unwilling to let Meta's bots crawl (automatically scan and copy) their linked content should stop using Meta services. Nieman Lab reported that the change covers all business accounts, and legal experts have raised alarms about it. Meta has not publicly commented on the terms.
News publishers have the most to lose. Their articles could command millions of dollars in licensing fees if Meta had to pay for continual access to train its models. Under the new terms, that leverage could shrink.
The timing matters. News organizations are currently fighting OpenAI over whether training AI on their work without a license counts as fair use, the legal doctrine that allows some unlicensed copying. A court ruling for publishers would make unlicensed training costly. A ruling for OpenAI would make licenses optional. Meta's terms seemingly risk granting it sweeping permissions no matter which way that fight goes.
That is the useful way to read the change. Copyright law asks what an AI company may take without asking. A contract can answer a different question: what a customer agreed to hand over in exchange for a service. If a publisher clicks accept, a later argument that Meta needed a license gets much harder to make.
The open question is how freely a business can refuse when the only alternative is leaving the platform entirely.