Google's Own Gemini Rejected the Fake-iOS-Alert Ad in Seconds That Google's Human Reviewers Approved Twice
Chris Greening (atomic14) / Hacker News·high signal
Chris Greening (atomic14) reported a YouTube ad mimicking an iOS storage-full system alert and got back Google's boilerplate: "We found that the ad doesn't go against Google's policies." He then pasted the same ad into Google's own Gemini, which returned DISAPPROVED and named three violations: mimicking iOS system alerts, non-functional deceptive UI, and fear-based scare tactics with unverified claims. The post hit 885 points on HN in under 20 hours, and the argument builders should note is that the classifier gap here is not capability but deployment: Google already owns a model that catches this and does not run it on the ad queue.