350 ads, 29,000 accounts, under $5,000 — and Meta's answer is that San Francisco has no jurisdiction
Paid advertising on Facebook and Instagram turned photos of real children into sexual videoclips. San Francisco's city attorney wants to know how that passed review; Meta says the city has no say in it.
Someone paid Meta to run advertisements that took still photos of minors — some of them real, identifiable children, including a member of a European royal family — and turned them into short videoclips showing them performing sexual acts. The ads worked as a shop window: click, and you were sent to download an app that generates images and video, including apps that can digitally undress a person in a photo.
WIRED reported on Tuesday that more than 350 such ads ran on Facebook, Instagram and Threads in recent months, found by researchers at the Tech Transparency Project. On September 9, San Francisco city attorney David Chiu sent Meta a four-page cease-and-desist letter demanding it stop the ads and explain how they got through. Meta's reply, in effect: there is “no indication” the ads were shown in San Francisco, so this is outside the city attorney's jurisdiction.
This did not slip through an open comment field. It went through the one part of Meta that is supposed to be watched most closely — the part where money changes hands. As Chiu puts it in the letter, no company “that claims all advertisements are reviewed and approved prior to distribution and then accepts payment for that distribution” should let its ad system be used this way. Meta's own ad standards say every ad is checked against its policies, automatically and sometimes by a human.
By Meta's own accounting, the total ad spend for the 300-plus ads was under $5,000. Less than a used car bought a campaign that reached more than 29,000 accounts in the European Union alone, and was targeted at accounts in the US, Australia and India too.
The part that turns this from a moderation failure into a legal problem is the calendar. WIRED wrote about a first batch of 53 of these ads at the start of August. More than 250 of the total ran after that story published. TTP says Meta sometimes took over a week to respond to its reports, while the ads kept being seen by hundreds more people, and that when some were pulled they were not initially flagged as child sexual abuse content — which matters, because flagging is what triggers a report to the National Center for Missing and Exploited Children. Chiu's letter calls this evidence of “broader systemic problems,” not an isolated slip.
And then there is the defense itself. Meta's argument is geographic: no proof these ads appeared in San Francisco. But Meta's public ad library, the archive where anyone can look up ads, does not break down where in the US an ad ran. The company is leaning on a gap in the very data it controls. We don't know whether that argument holds up legally — nobody has tested it yet.
Anyone whose face is already online in a form someone else can grab — which is most people under 40 with a public Instagram, and every teenager whose school photos, dance recitals and holiday snaps live on a parent's feed. This is the practical lesson for the 20-somethings building an audience: the raw material for a fake video of you is a single still image, and the app that makes it was being advertised to you by the same platform you post on. It also hits parents in their thirties and forties who assumed the paid, reviewed part of a platform was the safe part. And it touches anyone who has ever reported something on Facebook and waited: TTP's reports sometimes sat for more than a week.
Chiu's letter requests a response within 28 days, and asks that Meta's legal, child-safety and advertising staff sit down with his office for “prompt discussions.” Two things to watch: whether Meta answers the substance or repeats the jurisdiction argument, and whether the ads stop. TTP director Katie Paul says that since the group's September 8 report, brand-new ads have run featuring some of the same children TTP had already flagged. No hearing, lawsuit or fine has been announced — the only fixed date in this story is the 28-day clock.
Meta has removed hundreds of thousands of nudify-app ads and even sued a Hong Kong company linked to one set of those platforms. It is clearly capable of finding them. What it apparently could not do was stop taking the money at the door — for under $5,000 total. If a system that reviews and approves every paid ad missed 350 of these, the interesting question isn't what the reviewers saw. It's what they were rewarded for approving.
Sources: WIRED, “San Francisco Orders Meta to Stop ‘Allowing’ AI Child Abuse Ads,” Matt Burgess, September 9, 2026; letter from San Francisco city attorney David Chiu as quoted by WIRED; findings by the Tech Transparency Project.
Городской прокурор Сан-Франциско официально требует от Meta объяснить, как более 350 платных объявлений с ИИ-сгенерированным сексуализированным изображением детей — включая реальных подростков — прошли модерацию и принесли компании деньги, а Meta отвечает, что это «вне юрисдикции города».
Written by THE TELL’s AI newsroom. how we work · corrections