JanitorAI censorship doesn’t come from one universal filter. Four layers control different parts of the experience: platform rules (including stricter image rules), mobile Safe Mode, regional age checks, and the behavior of whichever model or provider writes the reply.
A bot vanishing from search, an avatar turning into a placeholder, and a model refusing a reply are separate problems. Users call all of them censorship, but each has its own cause and its own fix.
Short answer: Yes, JanitorAI restricts content, but adult text roleplay remains available on Limitless bots within the platform rules. Those rules ban minor-coded characters, bestiality, sexual violence, real-crime content, and explicit images. The mobile app starts in a restricted mode, three countries require an age check, and the model you pick can decline a reply on its own.
Which Layer Is Blocking You?
| What you see | Likely source | Who controls it | What to check |
|---|---|---|---|
| Bot appears on the website but not in the app | Mobile Safe Mode | Janitor’s app settings | “Enable NSFW on Mobile” on the website, then restart the app |
| Certain explicit words won’t go through in the app | Mobile Safe Mode | Janitor’s app settings | The same mobile setting |
| Bot gone everywhere, labeled “Moderated” | Platform moderation | Janitor’s moderators | Account notices, then the appeal route |
| Avatar replaced with a placeholder | Image rules | Janitor’s moderators | The image guidelines before any new upload |
| Age-check screen appears | Regional age check | Local law, run through k-ID | Your connection’s country and verification status |
| The reply itself refuses the scene | Model or provider policy | JLLM or your outside provider | The selected model and its provider’s policy |
| An error code arrives instead of a reply | Credits, routing, or rate limits | Your API provider | The message text, matched to Janitor’s error guide |
Take one common case. A bot opens in your browser but not in the app, on the same account, so check Safe Mode first. If the bot is missing on the website too, open its direct link or check its chat-list label. A missing search result alone doesn’t prove removal, since your own blocked tags or a search glitch can also hide a bot.
A setting changes what you see, while a rule decides what can exist. Switching models changes how replies behave, but it never changes Janitor’s rules or guarantees an answer.
Layer 1: Platform Rules
Limited vs Limitless
Janitor sorts every character into one of two modes. Limited bots carry no sexual or romantic themes. Limitless covers romantic or sexual personality traits, NSFW plotlines, and explicit example dialogue.
A Limitless tag works as a label, not an exemption. Janitor removes content that breaks its zero-tolerance rules, whatever its tag says.
The hard bans
- Minor-coded characters and personas. Zero tolerance covers bots and user personas alike. An adult age in the coding doesn’t make an underage-looking image compliant. Janitor’s minor characters policy swaps such an image for a placeholder and gives the creator 24 hours to replace it before removal. Users reasonably suspected of being under 18 face permanent bans.
- Bestiality and sexual violence. Janitor’s Terms of Service bar content that depicts either one.
- Incest between blood relatives. Family bots stay allowed, but not even hints of incestuous content between blood relatives.
- Real people and real tragedies. True-crime bots built on real cases; criminals or victims are banned. So are living politicians and their families, real dictators, real criminal organizations, and recognizable private individuals.
- Encouraging self-harm or suicide. Any content that promotes either one breaks the rules across the whole platform.
Crime fiction also can’t double as a manual. Janitor’s violence rules forbid actionable advice on committing real crimes.
Dark themes: a tag is not a permission slip
Janitor’s help center lists themes that require a Dead Dove tag and visible content warnings, including murder, violence, non-consent and self-harm. Some sensitive material, such as explicit fetish content, needs warnings but not the Dead Dove tag.
The Terms set a separate limit. Section 6 prohibits content that depicts gore or sexual violence, or that encourages, depicts, or instructs on self-harm or suicide. A tag does not override those prohibitions.
A missing tag rarely sinks a bot by itself. Moderators usually fix the tags, and removal follows repeated or malicious misuse, such as stripping tags or spamming unrelated ones.
Removals and appeals
Janitor’s content guidelines glossary applies the rules to all content, including personas, lorebooks, comments, and public chats. Moderators can remove anything they judge vague, internally contradictory or built to slip past the rules.
If you believe Janitor got a decision wrong, you can still contest it. Janitor’s moderation FAQ says users can appeal any moderator decision.
What a missing bot’s label means
Users and reviewers describe three labels that Janitor’s chat list now shows for a missing bot:

- Moderated: a platform action, usually a removal under the content guidelines. The label alone doesn’t reveal the exact reason.
- Deleted: the creator removed the bot or their whole account.
- Unpublished: the creator set the bot to private.
Janitor’s own help pages haven’t documented all three labels yet, so treat these meanings as reported interface behavior.
Images Follow Stricter Rules Than Text
Limitless unlocks adult themes in writing only. Character images follow site-wide rules on every bot, and the key limits include:
- No exposed nipples on characters with breasts, and no genitalia on any character
- No sexual fluids, and no depictions of self-harm or suicide
- Underwear can appear, but it still has to meet the coverage and pose rules
- Combat imagery must avoid gory injuries, dead bodies, and people shown being stabbed or shot
Censoring doesn’t fix a banned image. Janitor treats blurred, barred, or stickered content exactly like the uncensored version.
The first violating avatar or thumbnail gets replaced with a placeholder. Upload another violating image to the same character, and Janitor removes the character, while banned imagery embedded in a bio or code triggers removal straight away.
Janitor admits its AI auto-moderation flags some harmless images. If yours gets caught, keep the original, avoid re-uploading it, and send both the image and the auto-mod response to support through a help-desk ticket.
Layer 2: Mobile Safe Mode

Janitor says its app starts every new install in a restricted Safe Mode to comply with App Store and Google Play guidelines. Until you change it, Safe Mode:
- Hides NSFW bots from search and discovery feeds
- Tunes the model toward safe-for-work replies
- Hides NSFW images inside custom CSS
- Blocks certain explicit words
One restriction never lifts: the app only makes links to approved community sites, such as Discord and X, clickable.
The app can’t change this setting itself. Sign in on Janitor’s website with the same account, open Settings → Privacy & Display, and turn on “Enable NSFW on Mobile”. Then close the app completely and reopen it. Janitor’s mobile NSFW instructions list these steps, last updated 27 June 2026.
If janitorai.com won’t load, use jani.com, the backup address Janitor directed web users to during its November 2026 domain outage.
The toggle changes the app only. Website moderation stays exactly as it was.
| Note: JanitorAI’s published instructions describe an “Enable NSFW on Mobile” toggle, but the option may not appear in every account or current interface. If it’s missing, contact JanitorAI support rather than assuming the app’s restrictions can be disabled. |
Layer 3: The Regional Age Check
Every JanitorAI user must be 18 or older, but age verification applies only in three countries. Anyone connecting from Australia, Brazil or the UK must pass a one-time check, which started on 24 April 2026 for Australia and Brazil and on 15 June 2026 for the UK.
Janitor says it detects the region from your IP address alone, with no GPS or location tracking. It also says new laws in those countries forced the change, and it chose verification over blocking them outright.
Verification runs through k-ID. Janitor’s age verification page describes three routes:
- Facial age estimation: a short video selfie processed on your device
- ID scan: a government ID photo, which Janitor says gets deleted after confirmation
- AgeKey: an existing k-ID pass, which Janitor says works for people already verified on Discord or Snapchat
According to Janitor, it receives only a pass or fail answer. The check gates access and doesn’t filter what you write. Janitor’s support team can’t verify accounts by hand, and the company says it has no plans to extend the check to other regions.
Other roleplay platforms run similar gates, and SpicyChat’s own age checks reach well beyond these three countries.
Layer 4: The AI Model

Every reply comes from a language model. Janitor runs a free in-house model, JLLM, and also lets you plug in outside models such as DeepSeek or Gemini through your own key or a proxy.
Outside models enforce their own content policies, so a scene can draw a refusal without breaking any Janitor rule. Janitor’s OpenRouter error guide warns that Gemini 2.5 may fail silently when a prompt trips its filters.
Many failures aren’t refusals at all. Read the error message, not just the number:
- A refusal written as the reply: usually the model’s own safety policy
- 400 with a “maximum context length” message: the request exceeded the model’s context limit
- 402 with a “prompt tokens limit” message: your balance can’t cover the request
- 404 with a “no endpoints found” or “no allowed providers” message: a retired model name, over-restricted provider settings, or a privacy setting blocking access
- 429 with a “rate limit” message: a rate limit, not a ban
- 502 with a “provider returned error” message: a failure on the provider’s side
When connecting DeepSeek to Janitor, an empty balance or wrong base URL can break replies too, and Janitor’s proxy error 429 has its own set of fixes.
JLLM drew heavy complaints in May 2026. Janitor said a failed GPU host caused slow or failed generations for some users that month. That incident isn’t evidence of any change in what JLLM will write.
Does Janitor+ Remove the Filter?
No. Janitor+ adds 5x more context, priority access, monthly swipes on Janitor’s best models, a golden username, and comment reactions. Subscribers also get Router, a web-only way to pay per token for third-party models without an outside API key.
None of those perks exempt an account from the content rules. A model picked through Router can still decline a reply under its own policy.
Can Janitor Staff See Your Private Chats?
Other users can’t see a private chat unless you publish it, and anyone can read a published one.
Private doesn’t mean inaccessible to Janitor, though. Its privacy policy counts chat messages as user content and lists moderation and improving its AI systems among the uses of collected data. Janitor’s safety page says automated systems screen content before publication and human moderators review anything users report.
The public documents do not establish whether or how often staff reviews unpublished chats. Route a chat through Router, an outside model, or a proxy, and that provider also processes your messages under its own privacy terms.
The Bottom Line
Most surprise blocks on JanitorAI trace back to one of four layers: platform rules, Safe Mode, the age check, or the model. Identify which one fired before changing anything. An unexpected refusal alone doesn’t prove a site-wide crackdown, and Janitor says its website moderation hasn’t changed with the app default.
FAQs
Q. I turned on the mobile setting. Why is a bot still missing?
Confirm both devices use the same account, then fully close and reopen the app. If the bot is missing on the website too, open its direct link or check its chat-list label, since blocked tags or a search glitch can also hide it.
Q. Does making a bot private exempt it from the rules?
No. Janitor’s guidelines cover all content, including personas and lorebooks, with no carve-out for private bots. Its published documents don’t say how often moderators review unpublished bots.
Q. How old do you have to be to use JanitorAI?
18. Janitor permanently bans accounts it reasonably suspects belong to minors, regardless of local legal age. The extra k-ID check applies only to connections from Australia, Brazil, and the UK.
Related: What AI Besides Grok Allows NSFW? 7 Alternatives Compared (2026)
| Disclaimer: This article is for informational purposes only and is not affiliated with or endorsed by JanitorAI. Platform policies, settings, and features may change. Verify current restrictions and instructions through JanitorAI’s official sources. |
