← Overview | Case Study

A dangerous toy reaches half a million parents.
Then the ad quietly disappears.

Picture a counterfeit baby teething ring, made with lead-contaminated plastic, advertised to parents across the US. The ad runs for a day and a half, reaches half a million people, and then it is gone. There is no copy of it anywhere, no record of how many people saw it, and no way to find the families who bought the toy. This is a made-up product standing in for a very real problem, and further down this page are three documented cases that show every part of it has already happened.

The product

A five-dollar toy that should never have reached a child.

"SoftBite Baby Teething Ring, BPA Free, Pediatrician Recommended"

$4.99 Ships from Shenzhen Contains lead Never safety-tested Fake certifications

The toy is made from cheap PVC plastic with lead levels many times over the US limit. Everything about the listing is borrowed or invented: the product photos are stolen from a legitimate brand, the "BPA Free" and "CPSC Approved" badges are fabricated, and the glowing reviews were written to order. It ships straight from an overseas factory through a throwaway seller account, the kind that can be closed and reopened under a new name in an afternoon.

The campaign

What 36 hours looks like from the inside.

Monday, 8:00 AM · Campaign launch

The ad goes live on Meta and TikTok

The seller puts $2,000 behind a tightly targeted campaign, aimed at women aged 22 to 38 who follow baby products, new-mom and infant-care content in Texas, Florida and California. The ad shows a smiling baby chewing the colourful ring, under the line "Pediatrician-approved. Only $4.99 this week."

Budget: $2,000Target: new mothers, 22-38

Monday, 12:00 PM · 4 hours in

The algorithm finds its audience

Meta's optimisation notices the ad doing well with first-time mothers aged 25 to 32 and pushes more of the budget their way. Engagement climbs, because the product is cheap, the photos look trustworthy, and the fake safety badges do their job.

Reach: 85,000 peopleOrders: ~200

Tuesday, 6:00 AM · 22 hours in

Half a million parents have now seen it

By now the ad has reached more than 500,000 people. Thousands have clicked through and hundreds have ordered. The toys are already moving out of an overseas warehouse toward American homes. Nobody has tested them, and no regulator has any idea the campaign exists.

Reach: 500,000+ parentsOrders: ~2,400

Tuesday, 8:00 PM · 36 hours

The budget runs out and the ad stops

The $2,000 is spent, so the ad stops running. The seller has taken about $12,000 for a toy that cost thirty cents to make. The seller account is closed, and a fresh one will go up tomorrow under a different name.

Revenue: ~$12,000Seller account: deleted

Tuesday, 9:00 PM · Hour 37

And then there is nothing left to find

The ad drops out of Meta's Ad Library, and it was never in TikTok's, because TikTok runs no ad library in the US at all. The content isn't saved, the reach isn't recorded, and the targeting is gone. As far as any public record is concerned, the campaign never happened.

Ad content: gone Reach data: gone Targeting: gone Advertiser: gone

Two weeks later

A child ends up in hospital

A pediatrician in Houston finds elevated lead levels in a nine-month-old. The parent remembers buying a teething ring online but not where, and not which ad. The listing is already gone, and so is the seller. The ad that put this toy in front of half a million parents exists in no public record anyone can search.

The aftermath.

🇺🇸 What an investigator can find in the US

No record that the ad ever ran
No reach data, so was it a thousand people or a million?
No targeting details, so no way to know who was shown it
No advertiser named, so no way to know who paid
No way to warn the other parents who saw it
Nothing to build a product-recall case on

🇪🇺 What the same investigator would have in the EU

The full ad, archived for a year
Exact reach, with a demographic breakdown
The targeting: mothers, 22 to 38, in TX, FL and CA
The advertiser and the payer, both named
Enough data to trace and warn everyone who saw it
Evidence a regulator can actually act on

In the EU, every step above would be on the record.

Run the identical campaign in Europe and, under the Digital Services Act, it leaves a complete trail that anyone can look up.

🇺🇸 When the ad ends
It is taken straight out of the Ad Library. Nothing about its content, reach or targeting is kept, and the investigation stops before it starts.
🇪🇺 When the ad ends
It stays archived for a full year, with its content, reach, targeting and the identity of whoever paid, all searchable by anyone.
🇺🇸 When the child is hurt
The parent can't find the ad and regulators can't trace it. With no evidence the campaign ever ran, a recall is next to impossible.
🇪🇺 When the child is hurt
Regulators search the archive, find the campaign, name the advertiser and see exactly who was targeted. A recall can go out within days.
🇺🇸 On TikTok
There is no ad library at all. Even ads that are running right now can't be inspected from the outside.
🇪🇺 On TikTok
A full DSA-compliant library covers live and past ads, with targeting, reach and engagement all on the record.

The toy was invented. These next three cases were not.

Every part of it has already happened.

We made up the teething ring so the whole chain would sit on one page. Each case below is real, documented, and lines up with one beat of that story. The sources are linked, so you can check them yourself.

The fake safety badges are a real tactic

US Consumer Product Safety Commission · May 2026

The CPSC opened a national crackdown on counterfeit safety labels and imitation certification marks, the very same "CPSC Approved" and "BPA Free" badges that scammers fake, used to move dangerous, mostly foreign-made products into American homes through e-commerce. In the regulator's own words, bad actors "are increasingly using counterfeit certification marks to evade US safety requirements."

How it maps to the storyThe invented "Pediatrician-approved" and "CPSC Approved" badges on our listing aren't a dramatic flourish. They are a tactic the regulator has documented and named.

Real scam ads really do run for two days and vanish

Which? · ACCC v Meta · 2025

The consumer group Which? found Facebook ads that used a deepfaked video of finance expert Martin Lewis and a fake BBC page to push a crypto scam. The ads stayed live "for only two days", long enough to reach victims but short enough to leave no lasting public trace. Australia's competition regulator has taken Meta to Federal Court over the same kind of celebrity investment-scam ads.

How it maps to the storyThe "36 hours, then gone" timing in our scenario isn't for effect. This is roughly how long these campaigns run before they disappear.

When investigators went looking, the data was already gone

US House Intelligence Committee · 2016 to 2018

Russia's Internet Research Agency ran 3,519 Facebook ads seen by 11.4 million Americans. They became public only because Congress subpoenaed and released them years later. And when a researcher showed that Facebook had under-counted their reach, Facebook deleted the cached ad data analysts had been relying on, describing it as fixing "a bug."

How it maps to the story"Nothing saved, no record it ever ran" is the real aftermath of the most scrutinised ad-influence case in US history. Even a hostile foreign campaign left almost nothing behind.
For $2,000, someone can put a dangerous product into thousands of homes, and a day and a half later there is no public record that any of it ever happened.

Nothing here broke or slipped through a gap. This is the system doing exactly what it was built to do, which is to remember nothing.

See the platform transparency data →