In early November 2025, Reuters published internal Meta documents that finally put a number on something serious advertisers have suspected for years.
Meta projected that about 10% of its 2024 revenue — roughly $16 billion — came from ads promoting scams and banned goods: fraudulent ecommerce, illegal casinos, and prohibited medical products. Users were shown an estimated 15 billion “higher risk” scam ads on an average day.
For anyone running certified health on Meta, that last category is the one that matters. Prohibited medical products sit in the same enforcement bucket as the grey compounding shops and unlicensed telehealth funnels that made GLP-1 advertising a mess. The platform does not always tell a licensed pharmacy apart from that mess.
Bans became silent penalties
The documents say Meta bans advertisers only when automated systems are 95% certain an account is committing fraud. Below that threshold — accounts that look risky but are not definitively scams — Meta applies penalty bids. Suspicious advertisers pay more to win the same auction.
Clear scam, potentially blocked. Suspected risk, charged more to stay in the auction. From Meta's side, a ban removes spend. A tax keeps the revenue and adds friction.
From a legitimate GLP-1 or telehealth advertiser's side, the problem is that those systems are not precise. If your assets share surface signals with the accounts being taxed, you inherit the premium without a notice.
That is a shift from the old Facebook Ads era, when pages vanished overnight. Outright bans are rarer now. Most brands still have their original Business Manager. Enforcement did not stop. It stopped announcing itself. Appealed violations, refund spikes, old Page flags — they stay on the record. The assets survived. The trust often did not.
What makes a health brand “look risky”
Automated systems do not distinguish well between actual fraud and a certified clinic that happens to share signals with bad actors. High refund or chargeback rates. Negative post-purchase feedback. Policy violations you already won on appeal. Recycled or previously restricted assets. Erratic spend. Complaints about shipping, product quality, or support — which, in this category, often means “the medication did not arrive when the ad said it would,” or “the consult did not match the landing page.”
The more your advertising footprint resembles the accounts Meta is taxing, the more likely you are to pay the same premium. That shows up as CPMs with no obvious cause, campaigns that will not scale, reviews that take longer or reject for vague reasons, and performance that degrades gradually. Most buyers assume it is competition or “the algorithm.” These documents suggest it can be reputation-based pricing.
What you can actually do
You cannot change Meta's incentive to monetise risk. You can change whether your stack looks like the stack being taxed.
That starts with the account layer. A cold Business Manager in a restricted health vertical already looks closer to the penalty bucket than a platinum account that has been running certified health at volume. Creative, bid, and audience tweaks will not move a trust tax.
It also means not giving the classifier extra reasons to group you with prohibited medical ads: claims you cannot support, funnels that bait-and-switch the consult, or media running through assets with a history you cannot see.
The Reuters leak confirms the marketplace is not neutral. Trust and risk are priced. For a certified health brand, the penalty for looking like a risky advertiser is often not a clean ban. It is higher costs and worse delivery, with nothing in Ads Manager naming the cause.
