Researchers went through Meta's public ad library and found more than 50 image and video ads running across Facebook, Instagram, Messenger and Threads that violated Meta's policies on child sexual abuse material, with AI involved in the content. Some were still running the week the story broke.
Most of the coverage is about Meta's responsibility, and rightly so. But there are two quieter lessons for anyone paying to advertise on those apps.
The first is that automated ad review is not a seal of approval, in either direction. If a platform's system can miss more than 50 ads of that kind, it can also wrongly reject your perfectly ordinary ad for a dental practice or a landscaping business. Plenty of small business owners take a rejection personally, or assume they have done something wrong, and start rewriting good copy to please a machine. Usually the better move is to appeal, ask a human to look, and keep a record of what got rejected and why.
The second is adjacency. You do not control what appears above or below your ad in someone's feed, and you do not control what else the platform is running that day. That is the deal with rented space. It is a reason to make sure the traffic you buy lands somewhere you own, your own website, on your own domain, collecting your own email addresses. If a platform has a bad week, or your account gets suspended by mistake, your business should still be reachable.
Worth knowing too: the ad library that researchers used is public. Anyone can look up the ads a business is running, including yours and your competitors'. That is useful for research and slightly uncomfortable if you assumed your campaigns were private.
What to do
Nothing urgent. Next time you speak to whoever runs your ads, ask two things: which placements your ads are set to appear in, and whether you can narrow that list. Then check that every ad you run points to a page on your own site rather than a profile or a form inside the platform.
Reported by:
