Meta has paused the ad terms that would have put its AI's mistakes on you
New terms making advertisers answer for Meta's AI edits, and letting it crawl their websites, won't start on 20 November. Your ads can still be changed by AI, so use the breathing space to check them.

Meta has paused the new terms it wanted every business on its ad tools to accept. On Thursday 5 November it took down the preview and posted a notice in its Business Help Center: updates to its Commercial Terms, Product Catalog Terms, Self-Serve Ad Terms and Ad Creative Generative AI Terms "are now paused and will not take effect on November 20, 2026".
The terms were first due on 30 October, then pushed to 20 November. Now there's no date. A Meta spokesperson said the company wants to "take our time to get this right". For now, the notice says, "our existing terms remain in effect".
“The pause changes the paperwork, not what Meta's AI does to your ads.”
Two parts drew the fire. One would have let Meta crawl the websites linked to any business account and scrape content from them. The other, reported by Adweek, would have made advertisers "solely responsible" for whatever Meta's AI features create or change in their ads, including features Meta switches on itself. Meta's earlier wording said you "may choose" to use its AI ad tools, and were responsible for "your use of any Output". The new wording dropped the choice.
Why it matters
The backlash came mostly from news publishers, but the terms covered anyone who runs ads. The UK's Competition and Markets Authority opened a formal investigation on 30 October, covering terms for "publishers and advertisers". A Norwegian court had already blocked the terms temporarily. Meta calls this a pause and hasn't withdrawn the terms, so expect them back in some form.
The pause changes the paperwork, not what Meta's AI does to your ads. Adweek reported Meta's AI adding sitelinks, translated copy and AI-generated images to live ads, sometimes where the advertiser hadn't switched the feature on. Bookshop.org said one of its ads was "'enhanced' (or ruined, really) by Meta without our knowledge". The digital agency Exverus has had to pull live ads to strip out changes.
Meta says you can turn off individual Advantage+ features in Ads Manager at any time. The catch, said Exverus's Ryan Adams, is that many of the options sit in hidden or collapsed menus: "If every one of those isn't manually unchecked, the enhancements still apply."
If you run your own brand's ads, that's your name on an image you never approved. If you run ads for clients, it's worse: the client sees the ad, and you're the one who has to explain it.
What to do this week
There's no new date, so you've got breathing space you didn't expect. Spend an hour of it like this:
- Open every live campaign in Ads Manager and go through the Advantage+ creative options ad by ad. Untick anything you haven't chosen on purpose.
- Preview each ad in every placement. You're checking what people actually see, not what you uploaded.
- Screenshot the settings and date the file. If an AI edit ever causes a complaint, you'll want proof of what you'd switched on.
- If you're an agency, add one line to your client sign-off: which AI enhancements are on, and who approved them. It turns an awkward call into a routine one.
- Decide what Meta's bots can read on your site. Meta's crawler page lists meta-externalagent for uses such as training foundation AI models, and meta-externalads for improving its ad and business products. You can block either in robots.txt. Leave facebookexternalhit alone: it fetches the title, description and thumbnail people see when they share your links.
Then put a reminder in the diary for the new terms. When they come back, read the AI section before you click accept. If responsibility still doesn't depend on you switching a feature on, that's when to turn the automation off by default and switch on only what you've checked.
Also in this issue

Cold email benchmarks disagree by 4.6x
Instantly says the average cold email gets a 3.43% reply rate. Smartlead's typical sender gets 0.74%. If your 2027 plan rests on one of them, here's how to redo the maths with your own numbers.

What gets past LinkedIn's slop filter
LinkedIn checks every post it shows beyond your network for generic AI filler. Using AI isn't the problem. Having nothing to say is. Here's how to write posts only you could have written.

Trade deliverables, not discounts
A new survey of 200 US clients found 84% want the same price with better service, or more of it, once their provider uses AI. Here's how to answer the discount question at renewal.