Why is Google still serving dodgy ads?

为什么谷歌仍在投放虚假广告？

TLDR: AI is really good at detecting deceptive adverts - why isn’t Google using it? 简而言之：人工智能在检测欺诈性广告方面非常出色——为什么谷歌不使用它呢？

A couple of months ago I noticed this advert popping up in the YouTube app. The main reason I noticed it is that I accidentally clicked it - I knew I was running out of space on my iPhone and a minor lapse of concentration caught me out. These things can slip through even the best review processes, so I did my internet duty and reported the advert. 几个月前，我注意到 YouTube 应用中弹出了这则广告。我注意到它的主要原因是我不小心点开了它——当时我知道我的 iPhone 存储空间快满了，一时疏忽就中了招。即使是最好的审核流程也难免会有漏网之鱼，所以我尽了网民的义务，举报了这则广告。

The response I got back was less than satisfactory: Again, sometimes things slip through the net and people don’t always check things properly, so I reported it again. And got the same response. In fact, multiple people have reported this advert and received the same: We found that the ad doesn’t go against Google’s policies, which prohibit certain content and practices that we believe to be harmful to users and the overall online ecosystem. 我得到的回复并不令人满意：话说回来，有时确实会有漏网之鱼，人工审核也不一定总是那么严谨，所以我又举报了一次。结果收到了同样的回复。事实上，已经有多人举报过这则广告，但得到的回复如出一辙：“我们发现该广告并未违反谷歌的政策，我们的政策禁止我们认为对用户和整体在线生态系统有害的特定内容和行为。”

I’m a big fan of Hanlon’s razor. Never attribute to malice that which is adequately explained by stupidity. However, if you were being uncharitable, you do have to question whether it’s in Google’s interest to remove adverts that get lots of clicks. After all, these adverts are probably performing well and bringing in lots of lovely 🤑. A simpler explanation is that the people reviewing adverts cannot properly review everything so things get missed. You can’t help thinking that AI should come to the rescue here. Given Google has some pretty good models - surely it can put them to use? 我非常推崇“汉隆剃刀”原则：能用愚蠢解释的，就不要归咎于恶意。然而，如果你往坏处想，你不得不怀疑谷歌是否有动力去删除那些点击率极高的广告。毕竟，这些广告可能表现良好，并带来了丰厚的利润 🤑。一个更简单的解释是，负责审核广告的人员无法妥善审查所有内容，因此会有遗漏。你不禁会想，人工智能应该在这里派上用场。考虑到谷歌拥有一些相当出色的模型——难道不能利用起来吗？

Here’s what Google’s Gemini has to say about the advert: 以下是谷歌的 Gemini 对这则广告的评价：

Classification: DISAPPROVED 分类：未通过

Policy Violations: 政策违规：

• Misrepresentation: Misleading Ad Design (虚假陈述：误导性广告设计)
• Misrepresentation: Unreliable / Deceptive Claims (虚假陈述：不可靠/欺骗性声明)

Detailed Reasons for Disapproval 未通过的详细原因

1. Mimicking System Alerts / UI Elements (Misleading Ad Design) 1. 模仿系统警报/界面元素（误导性广告设计） The Violation: Google Ads policy strictly prohibits advertisements that imitate operating system dialogs, system warnings, error messages, or interactive system notifications. 违规内容：谷歌广告政策严格禁止模仿操作系统对话框、系统警告、错误消息或交互式系统通知的广告。 The Evidence: The banner embedded inside the ad explicitly mimics an iOS system alert modal (“iPhone Storage is Full”) complete with standard iOS typography, container styling, and mock system buttons (“Yes” / “No”). This trick visually deceives users into believing their device is throwing a native system-level alert, coercing them to click. 证据：广告中嵌入的横幅明确模仿了 iOS 系统警报弹窗（“iPhone 存储空间已满”），并配有标准的 iOS 字体、容器样式和模拟系统按钮（“是”/“否”）。这种伎俩在视觉上欺骗用户，让他们以为设备发出了原生的系统级警报，从而诱导他们点击。

2. Non-Functional / Deceptive UI Components 2. 无功能/欺骗性界面组件 The Violation: Ads cannot feature non-functional or misleading interactive elements, such as fake dialog options, false close buttons, or radio choices that do not perform their implied system function. 违规内容：广告不得包含无功能或误导性的交互元素，例如虚假对话选项、虚假关闭按钮或无法执行其隐含系统功能的单选框。 The Evidence: The “Yes” and “No” options inside the graphic image are static visuals designed to capture clicks anywhere on the ad banner to trigger a store/landing page redirect, rather than acting as true系统 controls. 证据：图片中的“是”和“否”选项只是静态视觉效果，旨在诱导用户点击广告横幅的任何位置以触发跳转到商店/落地页，而不是真正的系统控件。

3. Deceptive Fear-Based Tactics & Unverified Claims 3. 欺骗性的恐吓策略与未经证实的声明 The Violation: Advertising policies forbid using deceptive claims or scare tactics to induce panic and force immediate user action. 违规内容：广告政策禁止使用欺骗性声明或恐吓策略来诱发恐慌并强迫用户立即采取行动。 The Evidence: Stating “If you don’t free up space soon, some features may not work properly” inside a fake system popup fabricates a urgent state of failure on the user’s personal device, misleading them about their hardware status to drive app installs. 证据：在虚假的系统弹窗中声称“如果不尽快释放空间，某些功能可能无法正常工作”，这捏造了用户个人设备处于紧急故障状态的假象，误导用户对硬件状态的判断，从而诱导安装应用。

Required Action 所需操作 Disapprove the ad creative immediately. Issue a policy violation warning/disapproval flag to the advertiser account under Misrepresentation (Misleading Ad Design). Repeated violations of this nature may result in full account suspension for deceptive practices. 立即拒绝该广告素材。针对该广告主账户，以“虚假陈述（误导性广告设计）”为由发出政策违规警告/拒绝标记。此类违规行为若反复出现，可能导致账户因欺诈行为被永久封禁。

Google’s own model rejects the ad in seconds, yet Google’s review process approved it twice. Come on, meatbags - use some of the amazing AI tools you have access to. 谷歌自己的模型在几秒钟内就拒绝了这则广告，但谷歌的审核流程却两次批准了它。拜托了，人类同胞们——用用你们手头那些强大的人工智能工具吧。