Why is Google still serving dodgy ads?

TLDR:

AI is really bully astatine detecting deceptive adverts - why isn’t Google utilizing it?

A mates of months agone I noticed this advert popping up successful the YouTube app. The main logic I noticed it is that I accidentally clicked it - I knew I was moving retired of abstraction connected my iPhone and a insignificant lapse of attraction caught maine out.

These things tin gaffe done moreover the champion reappraisal processes, truthful I did my net work and reported the advert.

The consequence I sewage backmost was little than satisfactory:

Again, sometimes things gaffe done the nett and group don’t ever cheque things properly, truthful I reported it again. And sewage the aforesaid response.

In fact, aggregate group person reported this advert and received the same:

We recovered that the advertisement doesn’t spell against Google’s policies, which prohibit definite contented and practices that we judge to beryllium harmful to users and the wide online ecosystem.

I’m a large instrumentality of Hanlon’s razor.

Never property to malice that which is adequately explained by stupidity

However, if you were being uncharitable, you do person to mobility whether it’s successful Google’s liking to region adverts that get tons of clicks. After all, these adverts are astir apt performing good and bringing successful tons of beautiful 🤑.

A simpler mentation is that the group reviewing adverts cannot decently reappraisal everything truthful things get missed.

You can’t thief reasoning that AI should travel to the rescue here. Given Google has immoderate beautiful bully models - surely it tin put them to use?

Here’s what Google’s Gemini has to opportunity astir the advert:

---

Classification: DISAPPROVED

Policy Violations:

1. Misrepresentation: Misleading Ad Design
2. Misrepresentation: Unreliable / Deceptive Claims

Detailed Reasons for Disapproval

• The Violation: Google Ads argumentation strictly prohibits advertisements that imitate operating strategy dialogs, strategy warnings, correction messages, aliases interactive strategy notifications.
• The Evidence: The banner embedded wrong the advertisement explicitly mimics an iOS strategy alert modal (“iPhone Storage is Full”) complete pinch modular iOS typography, instrumentality styling, and mock strategy buttons (“Yes” / “No”). This instrumentality visually deceives users into believing their instrumentality is throwing a autochthonal system-level alert, coercing them to click.

2. Non-Functional / Deceptive UI Components

• The Violation: Ads cannot characteristic non-functional aliases misleading interactive elements, specified arsenic clone dialog options, mendacious adjacent buttons, aliases power choices that do not execute their implied strategy function.
• The Evidence: The “Yes” and “No” options wrong the schematic image are fixed visuals designed to seizure clicks anyplace connected the advertisement banner to trigger a store/landing page redirect, alternatively than acting arsenic existent strategy controls.

3. Deceptive Fear-Based Tactics & Unverified Claims

• The Violation: Advertising policies forbid utilizing deceptive claims aliases scare strategies to induce panic and unit contiguous personification action.
• The Evidence: Stating “If you don’t free up abstraction soon, immoderate features whitethorn not activity properly” wrong a clone strategy popup fabricates a urgent authorities of nonaccomplishment connected the user’s individual device, misleading them astir their hardware position to thrust app installs.

Required Action

Disapprove the advertisement imaginative immediately. Issue a argumentation usurpation warning/disapproval emblem to the advertiser relationship nether Misrepresentation (Misleading Ad Design). Repeated violations of this quality whitethorn consequence successful afloat relationship suspension for deceptive practices.

---

Google’s ain exemplary rejects the advertisement successful seconds, yet Google’s reappraisal process approved it twice.

Come on, meatbags - usage immoderate of the astonishing AI devices you person entree to.

Written by

Blog Logo

Chris Greening

---

Published 13 Sep 2026