A deceptive advertisement mimicking an iOS system alert has repeatedly slipped through Google’s ad review process, despite the company’s access to AI tools capable of identifying such violations.
The ad in question falsely warns users that their iPhone storage is full, using fake system notifications and interactive elements designed to deceive users into clicking through to an app store. Multiple users reported the advertisement to Google, but the company’s review process approved it both times, responding that the ad does not violate Google’s policies.
According to the source, when Google’s Gemini AI model was asked to evaluate the same advertisement, it immediately classified it as disapproved and identified multiple policy violations. The AI analysis noted that the ad mimics iOS system alerts and warning dialogs, uses non-functional UI components designed to trick clicks, and employs fear-based tactics to induce panic about device storage. Gemini determined the ad violates Google’s prohibitions on misleading ad design, misrepresentation, and deceptive claims.
The discrepancy raises questions about why Google’s established review process failed to catch violations that its own AI model identified within seconds. The source suggests two possible explanations: either the manual review process is insufficient to properly evaluate all advertisements, or there may be financial incentive to approve ads that generate significant click-through rates and revenue.
The author notes that AI detection of deceptive advertising is already well-established technology. Given that Google has access to capable AI models like Gemini, the continued approval of clearly deceptive ads appears preventable. The source concludes by suggesting Google should deploy its AI tools more aggressively in the ad review process, rather than relying solely on human reviewers who may miss violations or lack sufficient resources to thoroughly evaluate all submissions.
This case illustrates a gap between the capability of available AI systems and their actual deployment in Google’s content moderation infrastructure.
Key facts
- A deceptive iPhone storage warning ad passed Google’s review twice despite user complaints
- Google’s Gemini AI model identified the same ad as disapproved and flagged multiple policy violations within seconds
- The ad mimics iOS system alerts to trick users into clicking, violating Google’s policies on misleading design and deceptive claims
- Google’s manual review process approved the ad despite clear violations that AI could detect automatically