Forbidden Words to Search in Japan: What is behind "Kensaku Shite wa Ikenai Kotoba"?

There is a particular kind of curiosity that gets sharper the moment you are told not to act on it. In Japan that curiosity has been cultivated for years, around a phenomenon known as Kensaku Shite wa Ikenai Kotoba (検索してはいけない言葉), which translates roughly to "words you should not search". Across several large wikis and forums, long lists of search terms circulate with explicit warnings attached. Type them in anyway and you may stumble on harmless curiosities, but you may also land on material that disturbs, disgusts or lingers longer than you would like.

It is more than a handful of macabre anecdotes. The lists have become a fixed part of Japanese internet folklore, comparable to creepypasta, urban legends or ghost stories, except that the "ghosts" here appear in the form of search results. To understand why these words have built an entire subculture, you have to read them alongside Japanese forum culture, the quirks of search engines and the way recommendation algorithms amplify the unusual. The phenomenon sits at the crossroads of three older currents: the kaidan tradition of Japanese ghost stories, the anonymous bulletin-board culture born on 2channel, and the global rise of creepypasta as a literary format for sharing digital horror.

What is Kensaku Shite wa Ikenai Kotoba?

The phrase Kensaku Shite wa Ikenai Kotoba breaks into three pieces: kensaku (search), shite wa ikenai (something you should not do) and kotoba (word or language). Taken literally, it means "words you should not search". The subculture took shape in the 2000s, on Japanese forums such as 2channel (2ch) and its successor Futaba Channel, before spreading to wikis, blogs and video platforms. The early lists were not glossy. They were plain-text threads where regular users posted a phrase, a short description of what it allegedly led to, and a rough warning level. Other users replied with their own experiences, often anonymously, and the most discussed threads gradually crystallised into wiki entries.

Parallel phenomena exist in the Western internet, from the English-language subreddit "Don't Search This" to scattered lists of "cursed search terms" on Tumblr and TikTok. The Japanese version, however, is older, more tightly curated and more closely tied to forum and wiki culture. Many wikis maintain entries for years, adding background stories, warnings, classifications and pointers to related terms. That long-term stewardship is part of what gives the Japanese lists their texture. A typical entry is closer to a small encyclopedia article than to a one-line scare, and the best of them read like a collaboration between a folklorist and a sceptical sysadmin.

The Japanese wiki of forbidden words

The most common format is a curated wiki where users collectively collect, describe and rate terms. A typical entry follows this structure:

• Term: the word or phrase that allegedly returns problematic results.
• Danger level: a rough rating from harmless to extremely disturbing.
• Description: what users typically encounter when they search, without spoiling too much.
• Background: forum debates, real incidents, rumours or urban legends that grew up around the term.
• Related terms: other words with similar risk profile or shared origin.

Unlike classic creepypasta, the point is not to invent a fictional horror. The entries tend to point to real or at least real-sounding artefacts on the open web: disturbing images and videos, anonymous forum threads, allegedly leaked files, content that lives on obscure domains. That mix of "it might be real" and "you should not check" is exactly what gives the lists their pull. Readers who would never willingly browse the deep web are tempted to run a quick search, partly because the term is right there, in plain text, on a wiki that looks almost academic. The format has been replicated in English, Spanish and Portuguese, usually with shorter entries and a faster turnover, but the same skeleton: term, rating, description, lore.

The wikis also carry their own meta-rules. Most of them include a top-page disclaimer that the project is informational, that visitors search at their own risk and that the goal is to document a piece of internet culture rather than to drive traffic to harmful material. The more reputable wikis moderate new entries carefully, prune low-effort additions and sometimes retire terms that have become too easy to weaponise. That curatorial layer is one of the things that sets the Japanese tradition apart from a casual "scary search terms" video on YouTube, and it is also why some entries have been quietly removed over the years as platforms cleaned up the underlying material.

How does the danger classification work?

The community sorts terms into danger levels. The exact scale varies from wiki to wiki, but it usually follows a similar pattern:

1. Level 1-2 (Curiosity): harmless urban myths, odd images or confusing search results.
2. Level 3-4 (Discomfort): content that can trigger nausea or lasting unease.
3. Level 5-6 (Trauma): real violence, disturbing audio files, intense phobias.
4. Level 7-8 (Terminal danger): illegal material, malware or content that most platforms remove on sight.

It is worth flagging that these levels are not an official risk rating. They function as a kind of community warning code, and the same term can sit at different levels on different wikis, depending on who wrote the entry and when. Anyone who still chooses to search does so on their own responsibility, and many wikis underline that the goal is awareness, not promotion of the material itself. Think of the scale less as a precise measure and more as a shared shorthand for "you probably do not want to see this". A few wikis have also started adding a separate flag for terms that are mostly hoaxes, which is a useful reminder that not every entry is meant to be taken at face value.

Well-known examples and their background

Naming specific search terms would be reckless. That is precisely the bait the lists trade on, and the wikis themselves warn against it. A more useful angle is to look at the categories that keep reappearing across the lists, because the categories are what most readers are actually curious about:

• Urban legends in search form: alleged evidence of ghost stories, unexplained phenomena or modern myths that began in schoolyards and migrated to forums. These entries usually point to long-running kaidan threads, sometimes decades old, and treat the search term as the modern incarnation of a story that used to be told around a campfire.
• Disturbing real-world images: accident footage, crime-scene photos or amateur recordings from crisis zones that spread through anonymous uploads. The danger level here is rarely about shock value on its own; it is about the fact that the images were never meant to be seen outside their original context.
• Psychological puzzles: images, audio clips or text passages engineered to provoke discomfort, fear or disgust, often through visual patterns, particular pitches or unresolved contradictions. This category borrows heavily from Western creepypasta, with roots in horror fiction by authors such as Junji Ito and the broader Japanese tradition of junji-kyofu (純粹恐怖), or "pure horror".
• Illegal content and malware: corners of the web where search terms act as bait, funnelling users into malware drops or dubious forums. The horror in this category is practical: a single careless search can end with a compromised device or a law-enforcement notice.

What ties most of these categories together is the fact that they sit at the seam between curiosity and self-protection. If you have ever considered typing one of these words and then stopped, you already know exactly why that seam matters. The wikis know it too, which is why the most thoughtful entries are written as warnings first and stories second. A useful side effect is that the categories map almost perfectly onto the structure of urban-legend research in folklore studies, where the same archetypes show up across very different cultures and centuries.

Why are we drawn to the forbidden?

The pull of forbidden or dangerous content is not new. As far back as Romanticism, ruins, dark forests and secrets counted as sources of aesthetic experience, and psychoanalysis has a word for it: the fascinosum, the appeal of the threatening. On the internet these effects intensify through anonymity, algorithmic recommendations and the illusion that anything is one click away. A search engine flattens the cost of looking. There is no locked door, no librarian, no social signal that the people around you know what you are about to read. That absence of friction is itself a kind of invitation, and recommendation systems make it worse by surfacing adjacent material before you have decided to stop.

In Japan an extra layer comes from the long tradition of kaidan (怪談), the ghost and horror stories that have been passed on orally, in literature and, more recently, online. Many of these stories work because something unspeakable is hinted at and never fully resolved. The forbidden search words follow a similar logic. They promise knowledge you would probably rather not have, and that tension is exactly what keeps people reading the lists even when they have no intention of searching. The lists also borrow the structure of hyaku-monogatari (百物語), the Edo-period game of one hundred ghost stories, in which each new tale is supposed to summon something a little more real than the last. The format survives because it works on the same nerves the older stories did.

Legal situation in Japan

Japan regulates problematic content primarily through the Act on Limitation of Liability of Providers and Right to Demand Disclosure of Identification Information of the Senders (プロバイダ責任制限法), supported by measures from the Ministry of Internal Affairs and Communications (MIC) and voluntary filtering solutions. The MIC also runs awareness campaigns aimed at younger users, and most major Japanese ISPs ship with optional filters that can be switched on at the router level. Violations of existing law, from child sexual abuse material to extreme violence or incitement, are prosecuted regardless of whether the content is reachable through a search engine.

For users, the practical takeaway is straightforward. If you encounter material that breaks the law, in Japan or anywhere else, report it rather than redistribute it. The existence of a "forbidden list" does not relieve platforms or consumers of that responsibility. Japanese courts have repeatedly treated anonymous re-uploads of harmful material as a publishing act, not a private one, and the same logic tends to apply in most jurisdictions that host the wikis in the first place. Search engines, for their part, operate takedown pipelines that respond to legal notices, and the more reputable wikis will quietly retire entries that point to material which has been removed at the source.

How to engage with the phenomenon

Anyone interested in the cultural side of the topic can take a lot away from the wikis without ever running a risky search. A clear-headed look at three questions helps: Who added the term to the list? An editorially maintained wiki is not the same thing as an anonymous 4chan post. What is supposed to happen, exactly? Concrete descriptions read more seriously than vague threats. Which sources are cited? References to known media, research or official bodies raise the credibility of the entry, while terms that exist only as "trust me, do not search this" usually do not survive close reading.

A few habits go a long way. Treat the entries the way you would treat a well-built museum exhibit: read the wall text, look at the photographs that are clearly framed as documentation, and skip the rooms that advertise themselves as forbidden. If a term genuinely interests you as a research subject, search for academic writing, journalism and book-length studies on Japanese internet folklore first. The story is usually more interesting than the search result, and it does not come with the same risk profile. There is also a growing secondary literature in English on Japanese online subcultures, including translations and analyses of kaidan and creepypasta, which gives curious readers a safer on-ramp than the wikis themselves.

It is also worth watching the phenomenon evolve. The wikis predate social media, but TikTok and YouTube have added a new layer, in the form of creators who record themselves "reacting" to forbidden search terms in real time. Most of these videos are stitched together, heavily edited and not actually what the search returned, which is itself a small lesson about how the subculture reproduces. The best reactions treat the entries as folklore, not as instructions, and that distinction is the clearest signal that the content has crossed from document into performance.

In the end the phenomenon says more about us than about what is actually lurking on the web. The lists work because they tap into a basic human need to know what is behind the next door, even when everyone warns against opening it. Once you understand that, you can enjoy the fascination without being carried away by it, the same way you would read a good horror novel and then close the cover before bed.