7 min read
What Is Cloaking in SEO? Risks and How to Spot Accidental Cloaking
Cloaking means showing search engines one thing and visitors another. It breaks Google's spam policies, yet honest sites sometimes trigger the same pattern through hacked plugins, geo redirects or rendering setups. Here is how to recognise and rule it out.
Cloaking is one of the oldest black-hat SEO techniques and one of the clearest violations of Google's rules: the search engine is shown one page and visitors are shown another. Most legitimate businesses never set out to use SEO cloaking. The risk for them is different: a hacked site, a misconfigured server or an over-eager redirect can create the same mismatch without anyone noticing. This article explains what cloaking is, why it is penalised and how to check that your own site is not doing it by accident.
What is cloaking in SEO?
Cloaking is the practice of presenting different content to search engines and to users, so that a page ranks for something other than what visitors actually get. Google lists it as a violation of its spam policies.
Google's own examples include showing search engines a page about travel destinations while users see a page about discount drugs, and inserting text or keywords only when the visitor is a search engine. The defining feature is deception: the version used to rank the page is not the version people receive.
Cloaking is about substance, not presentation, and that separates it from legitimate tailoring. A page that rearranges its layout for phones, translates for a chosen language or shows a paywall to non-subscribers is not cloaking, as long as search engines and users receive the same substance.
How are search engines shown a different page?
Cloaking relies on a rule that decides who is asking for a page and returns a different version to crawlers than to people. The rule can use several signals, which is also why accidental mismatches happen.
The signals involved, described here so you know what to look for in your own setup:
- User agent. The browser identification string, which for Google's crawler includes "Googlebot".
- IP address. Known crawler address ranges, which Google and other operators publish.
- Location. Content that depends on the visitor's country or city.
- Rendering differences. Scripts that change the page after it loads, so the source HTML and what users see diverge.
- Request headers and DNS setups. Rules based on headers, or subdomains pointed at servers the owner does not control.
Software sold as a "Google cloaker" automates these rules for spam and fraud. Using it is a policy violation, not a grey area.
Why does Google treat it as spam?
Google penalises cloaking because it breaks the basic promise of a search result: that the page you click is the page Google evaluated. Sites that cloak may rank lower or disappear from results entirely.
Google detects spam through automated systems and, where needed, human review that can lead to a manual action. Manual actions appear in the Manual Actions report in Search Console, and a site has to fix the problem and request a review before the action is lifted.
In May 2026 Google clarified that its spam policies also apply to generative AI responses in Search. Content that tries to deceive the systems behind AI features is treated the same way as content aimed at classic results.
How honest sites end up with a mismatch
Accidental cloaking happens when a technical setup serves crawlers substantially different content from what users see, without any intent to deceive. The most common causes are hacks, inconsistent dynamic serving and location rules.
- Hacked sites. Google notes that attackers often use cloaking to hide injected spam from the site owner. The owner sees a normal page; Googlebot sees pharmacy or casino links.
- Dynamic serving drift. When mobile and desktop HTML come from different templates, one version can lose text, links or markup over time.
- Dynamic rendering and prerendering. Google calls dynamic rendering a workaround rather than a recommended solution. It is not cloaking while the rendered version is similar to what users see, but a broken prerender service can serve empty or outdated pages to crawlers.
- Geo redirects. Googlebot's default IP addresses appear to be in the USA. If US visitors are redirected to a different site version, Googlebot may never see the others. Google advises treating Googlebot like any other visitor from that country.
- Bot protection. Firewalls and CDN rules can serve challenge pages or errors to crawlers while people browse normally.
Paywalls are a special case. Google does not treat gated content as cloaking if it can see the full content that subscribers see, and paywalled content structured data helps it tell the two apart.
How to detect cloaking issues on your site
To detect cloaking, compare what Google receives with what a normal visitor sees, using Search Console, a crawler with different user agents and your server logs. Any difference in main content needs an explanation.
Website cloaking introduced by a hack can sit unnoticed for months, so these checks belong in every regular SEO audit, not only in a crisis.
| Check | What it reveals |
|---|---|
| URL Inspection live test in Search Console | The HTML, HTTP headers and screenshot Google's testing tool received, to compare with the page in your browser |
| Crawl with a Googlebot smartphone user agent and with a browser user agent | Differences in titles, text, links or redirects between the two crawls |
| Server logs with verified Googlebot requests | Status codes and redirects real crawlers received; verify them by reverse DNS or Google's published IP lists |
| Security issues and Manual actions reports | Whether Google has detected hacked content or issued a cloaking action |
| A site: search for your domain | Titles or snippets in results that do not exist on your pages, a common sign of a hack |
One caveat: the URL Inspection tool identifies itself as Google-InspectionTool rather than Googlebot. Server rules keyed strictly to the Googlebot string may treat it differently, so confirm findings with log data. If you find injected content, clean the site, close the vulnerability and then request a review.
Dynamic serving and prerendering: where is the line?
Dynamic serving and prerendering are legitimate when users and crawlers get the same main content in different technical forms. They become cloaking only when the substance differs: another topic, hidden keywords or a different destination.
A simple test helps: if a visitor could see the version served to Googlebot, would they feel misled? Different layouts for phones, prerendered HTML of the same article or a paywall with proper markup pass that test. Keyword blocks visible only to bots, or a crawler-only page that redirects people elsewhere, do not.
What about AI crawlers?
AI crawlers also identify themselves, and serving them different content from what users see is deceptive in the same way. Blocking them in robots.txt, by contrast, is a legitimate access choice, not cloaking.
OpenAI documents separate agents such as OAI-SearchBot for ChatGPT search and GPTBot for model training, with published IP lists. Anthropic documents ClaudeBot and related agents. Google-Extended is not a separate crawler but a robots.txt token that controls use of Google-crawled content for some Gemini purposes. Treat all of them the same way you treat Googlebot: either allow access to the real page or decline access openly.
Checks for rendering, dynamic serving and crawler access are part of QIO's technical SEO service.
Sources
- Spam policies for Google web search | Google Search Central
- Dynamic rendering as a workaround | Google Search Central
- Verify Requests from Google Crawlers and Fetchers | Google Crawling Infrastructure
- How Google Crawls Locale-Adaptive Pages | Google Search Central
- Overview of OpenAI Crawlers | OpenAI API
Questions and answers
What is an example of cloaking?
A page that shows search engines an article about travel while visitors are sent to an online pharmacy is a classic example taken from Google's own spam policies.
Is showing different content on mobile cloaking?
Not if both versions carry the same main content. Different layouts and device-specific HTML are normal; removing or replacing content only for crawlers is the problem.
Is dynamic rendering considered cloaking?
Google says dynamic rendering is not treated as cloaking as long as it produces similar content. It still recommends server-side rendering, static rendering or hydration instead.
Can my site be cloaking without my knowledge?
Yes. Hacked sites often hide spam from their owners using cloaking, and misconfigured redirects or bot rules can create similar mismatches.
Is blocking AI crawlers cloaking?
No. Refusing access through robots.txt is an open choice. Cloaking means serving a crawler different content while pretending it is the same page users see.
Follow the journal
New pieces on websites, SEO and AI come out in the QIO journal. Follow in Google, by RSS or in Telegram to get them first.
