Meta has quietly rewritten its Commercial Terms, and the new wording could let it scrape and train AI on news sites without paying a cent. The Verge AI reports that the updated terms give Meta permission to crawl any domain linked to a business that uses Meta’s products, then use that content to “train” its systems. If you’re a publisher sharing links on Facebook or Instagram, this one’s about you.
📌 The Short Version
- What changed: Meta’s Commercial Terms now include a broad clause that lets it crawl, cache and collect content from business websites.
- Who’s affected: Any company, publishers included, whose domains are “associated with” its products or come through Meta’s services.
- Why it matters: Meta may use that content to “provide, develop, improve, research, train” its products. That reads like permission to use it as AI training data.
- The big question: Whether agreeing to Meta’s business terms now means handing over your archive for free.
🔍 What the Terms Actually Say
The clause is broad. According to the text The Verge AI highlighted, businesses “authorize” Meta, or “any entity acting on our behalf,” to use “web scrapers, bots, robots, spiders, crawlers, and user agents” to access their sites.
It doesn’t stop at a single visit. Meta can “periodically pull updates” from those pages. It can even “create test credentials,” which suggests access that goes beyond public pages.
Then comes the key line. Meta “may use any content from such domains” to build and train “the Meta Company Products, services, and features.” No license fee is mentioned. No opt-out is spelled out in the excerpt either.
🧭 Why This Is a Big Deal
What stands out here is the approach. Instead of negotiating with publishers one by one, Meta wrote consent into the paperwork businesses already sign to run ads, manage pages or use its business tools.
That’s a sharp contrast with how the rest of the industry has handled it:
- Licensing deals: OpenAI, Google and others have signed paid content agreements with major news organizations to get training and display rights.
- Lawsuits: Plenty of publishers have sued AI companies over scraping without permission.
- Technical blocks: Many news sites use robots.txt rules to keep AI crawlers out.
Meta’s terms could sidestep all three. If you’ve agreed to the terms, you’ve arguably already said yes. That makes a robots.txt block harder to rely on, and it gives Meta a contract-based argument if things end up in court.
⚖️ Where It Gets Murky
The Verge AI frames this as something the terms “may allow,” and that hedge matters. A few things are still unclear:
- Scope: “Associated with your products or services” is vague. Does sharing one article link count? An ad account?
- Enforcement: Contract terms don’t automatically override copyright law. Publishers may push back.
- Regulation: In Europe especially, AI training rules and opt-out rights could clash with this kind of blanket consent.
Expect lawyers on both sides to fight over that language.
🛠️ What Publishers and Businesses Should Do Now
If your company uses Meta’s commercial tools, don’t wait for this to settle itself.
- Read the updated Commercial Terms. Pay close attention to the crawling and training clauses.
- Check your accounts. Work out which domains Meta could reasonably call “associated” with your business.
- Talk to legal counsel. Ask whether continued use counts as consent and whether you can negotiate an exception.
- Watch your server logs. Track Meta’s crawler traffic so you know what’s actually being collected.
- Coordinate with industry groups. Publisher trade groups will likely respond together, and that gives you more leverage.
🔮 What Comes Next
This is the clearest sign yet that the fight over AI training data is moving from robots.txt files and courtrooms into terms of service. Meta is betting that businesses need its platforms more than they’ll object to the fine print.
If it works, other platforms will probably copy the move. If publishers push back hard enough, Meta could be forced to narrow the clause or start paying the way some of its competitors already do. Either way, the free-scraping era just got a new legal wrapper. The original report at The Verge AI has the full terms text and more detail.