Skip to content

Product1 publisher3 min readPublished

San Francisco demands Meta account for 350 paid ads that cleared its own review

David Chiu's four-page letter asks Meta how the ads got past moderation, how it handles repeat advertisers and what it sent to NCMEC. That turns a takedown story into a question about what Meta owed once it knew.

The Product Desk · Product desk

Illustration accompanying San Francisco demands Meta account for 350 paid ads that cleared its own review

What happened

  • San Francisco city attorney David Chiu sent Meta's lawyers a four-page cease-and-desist letter ordering the company to stop allowing paid ads containing AI-generated child sexual abuse content.
  • WIRED reported that Meta ran more than 350 such ads in recent months, turning still images of minors, including a member of a European royal family, into clips that can depict them performing sexual acts.
  • The Tech Transparency Project, which found the ads, counted more than 29,000 accounts reached in European Union countries.
  • More than 250 of the ads ran after WIRED's early-August report on an initial batch of 53, according to WIRED's count.

Compiled by The Product DeskSomething wrong?How this is made

Why it matters

  • precedent A municipal law office, not a national regulator, is the venue that got a platform asked to explain its ad-review internals in writing, and four pages of demands is a cheap instrument to copy.
  • decision Any platform whose sales material promises that ads are reviewed and approved before distribution now has to decide whether that line is auditable per ad or should quietly come off the page.
  • exposure The demand for how Meta handles the advertisers and repeat offenders reaches past individual creatives to the accounts buying placement for nudify apps, which are the part of the operation that survives a takedown.
  • contradiction Meta's numbers describe cost and limited reach; Chiu's letter tracks what kept happening after he was notified, a different basis for judging the response.

The Tech Transparency Project said Meta went more than a week without answering some of its reports about these ads, and in that window the ads were seen by hundreds more people [20].

The money explains the persistence. Meta put total ad spend for the 300-plus ads at under $5,000 [16], which across the 350-plus creatives is under $15 apiece [1], and under 17 cents for each EU account the researchers counted [2] before you add the accounts reached in the US, Australia and India [6]. Low spend reads as reassurance in a statement. On an ad platform it is the reason the pattern repeats, because an advertiser at that price can lose every creative to eventual removal and still bank the app installs the ads were pointing at [3].

Meta's other numbers are reach numbers: many ads were pulled before TTP reported them, all have now been removed for policy violations, and most had fewer than 200 impressions [14]. Impressions describe delivery. Chiu's letter asks about the system that sold the delivery, demanding that Meta stop the ads, explain how they avoided moderation, explain how it escalates and reports ads to the National Center for Missing and Exploited Children, and detail how it deals with the advertisers behind them and with repeat offenders [17]. Only the first of those four is about the ads.

The re-upload detail is what makes repeat offenders the operative phrase. TTP found identical ads uploaded multiple times [4], and on WIRED's figures roughly seven in ten of the ads ran after its first story [3]. Chiu's letter puts it as findings that "do not appear to reflect an isolated failure that Meta corrected when first alerted to the problem" [11], and says they point to "broader systemic problems" in how the company detects child sexual abuse in advertising [18].

Meta's first line of defence is geographic: no indication the ads were displayed in San Francisco, so they sit outside the city attorney's jurisdiction [12]. The public record cannot settle that, because Meta's own ad library carries some US data but does not break down where in the US an ad ran [13]. The logs that would answer it are internal.

The lever Chiu picks up instead is the sell side. No company, he writes, "particularly one that claims all advertisements are reviewed and approved prior to distribution and then accepts payment for that distribution," should permit its advertising systems to be used this way [9]. Review is part of what an advertiser is buying, and the letter treats it as a representation rather than a courtesy.

For a platform where money buys distribution, two measurements and one wiring question follow. The first is hours from receipt of a third-party report to takedown, not from the moment a human opened the ticket. The second is the share of removals that are re-uploads of assets already removed, which is the number that separates a backlog from a hole. The wiring question is whether the removal path writes to the reporting path, because "removed for violating our policies" [14] and referred to NCMEC [17] are different records, and this letter asks for the second.

A cease-and-desist asks Meta's legal, child-safety and advertising staff for prompt discussions with the city office [10], though it carries no penalty by itself. It leaves behind a dated record of notice. Chiu says his objective is to ensure Meta "reliably satisfies the legal obligations that arise when it becomes aware of apparent child sexual exploitation" [19], which is the sentence that moves the argument off the moderation queue.

What to watch

  • Whether Meta's legal, child-safety and advertising staff take the meeting Chiu asked for, and what follows the letter if they do not.
  • Whether Meta produces delivery geography or NCMEC referral counts that would settle the jurisdiction objection on the record.
  • Whether TTP finds fresh ads dated after the letter, which is the direct test of the isolated-failure claim.
Loading claim ledger
Loading source directory links
Loading share composer
Loading topic controls
Loading related stories