Is Janitor AI safe?

Whether Janitor AI is safe depends on your age and on what you put into it. It is a platform where people create AI characters and chat with them, listed on the App Store as interactive stories, and it is for adults: the Terms of Service, last updated October 8, 2026, set the minimum age at 18, and the iPhone app shows an 18+ rating. The privacy policy, last updated September 17, 2026, lists what it collects and describes account deletion, but it gives no retention period and doesn’t say whether chats train models. Age verification through a provider called k-ID applies to users connecting from Australia, Brazil and the UK, and elsewhere the 18+ rule is a term you agree to and a reason to ban accounts the company suspects belong to minors.

We read the company’s own legal pages, safety pages, help center, newsroom and App Store listing on October 16, 2026, and used third-party write-ups only to find questions worth checking, apart from one disclosure rating that we cite below. In the sources we reviewed we found no regulator action, lawsuit or data breach involving Janitor AI.

Safety at a glance

The table below gives the documents’ answers in short form. Each row is explained in the sections that follow.

What Janitor AI is and who runs it

Janitor AI is published by JanitorAI Inc.

Janitor AI is a platform where people create characters and chat with them. The App Store listing describes it as a place to step inside stories, and files it under Books, with romantasy, fanfiction and interactive stories as its themes. That makes it a different kind of product from a single-companion app. The community guidelines describe millions of people creating and interacting with AI characters, so the quality, tone and content of any given chat depends on who made the character and how the platform moderates it. Our guide to AI girlfriend roleplay covers how that style of use differs from a continuing relationship with one person, and the Character.AI safety guide covers a platform of the same kind.

Its Terms name JanitorAI Inc. as the company, and its privacy policy says payments are processed by JanitorAI Inc. and its affiliates, including Dulova Investments Limited in Nicosia, Cyprus, as well as third-party processors. The Terms give support@janitorai.com as the general contact and list separate addresses for billing, privacy, legal and copyright matters.

Is Janitor AI safe for kids?

No. The Terms of Service say the platform is intended solely for users who are at least 18, that you represent you are 18 when you use it, and that you may not create an account or take part in its communities if you are under 18. The iPhone app is rated 18+ in the US App Store with in-app controls. The community guidelines call protecting minors the company’s highest priority, and its help center says users reasonably suspected of being under 18 are permanently banned.

The same documents say adult sexual content between consenting adult characters is allowed. Content with sexual or romantic themes has to carry the Limitless tag, and content tagged Limited is meant for general audiences. The company also bans characters under 18, characters whose age is unclear, and sexual or romantic content involving minors. A help article updated August 21, 2026 says characters must be explicitly defined as adults in the personality or scenario section, and that an age mentioned only in a title, tag or first message doesn’t count. Age play between consenting adults is allowed, and age regression is prohibited.

On iPhone, Janitor’s February 2026 post says the app starts in a stricter Restricted Mode because of Apple’s content rules, and that adult content is switched on from the website settings.

Janitor AI age verification

Janitor’s help center says users connecting from IP addresses in Australia, Brazil or the UK must complete a one-time age verification, and that users anywhere else see no change. The article, updated June 15, 2026, names three ways to do it: facial age estimation, an ID scan, or AgeKey, which reuses a verification you already completed elsewhere. The provider is k-ID. According to the company, k-ID doesn’t keep documents or facial data after verification, and Janitor receives only a yes or no on whether you are over 18. The company says the system relies on your IP address and doesn’t use GPS or ask for an address, and that existing characters, chats and settings stay as they were. The article also says there are no plans to extend verification to other regions.

Janitor’s June 15, 2026 newsroom post explains the reason. It cites Australia’s Age-Restricted Material Codes, in effect since March 9, 2026, and Brazil’s Digital ECA, in effect since March 17, 2026, and says the choice was between verifying users and blocking those countries or risking fines. The company says it opposes identity verification for social media and describes the age check as a requirement placed on it by those laws. In July 2025 it had blocked UK access, citing the UK Online Safety Act, and a 2026 help article lists UK users in the verification scheme, so access appears to have returned in some form. We found no post that says when or how.

Two details differ between the company’s own pages. The Safety page names Australia and Brazil as the verification regions, while the help article adds the UK. The help article also gives April 24, 2026 as the rollout date for Australia and Brazil, and the June 15 newsroom post gives no rollout date. If you are in one of these countries, expect the check; if you are somewhere else, the documents describe no age check beyond the 18+ representation in the Terms and bans for suspected minors.

Janitor AI privacy policy

The privacy policy was last updated September 17, 2026. It says Janitor collects what you give it directly: email address, username, password and profile details, the characters you create, your chat conversations and messages, payment information handled by its payment processors, and what you send to support. It also collects automatically: pages visited, features used, search queries, IP address, device information, approximate location derived from your IP address, and server logs. It uses cookies and similar technologies, and it says it doesn’t respond to Do Not Track signals.

If you sign in with Google, the policy says the sign-in runs through an authentication provider called Supabase and that Google supplies your account identifier, email address, name and profile picture. It says Janitor doesn’t receive your Google password and gets no access to your Gmail, Drive, contacts or calendar.

Training, sharing and sale

The policy lists several uses of your information, including content moderation, safety, analytics, and “conducting research and development to improve our AI systems and user experience.” That last line is the closest the policy comes to the training question. In the sections we read there is no sentence saying whether your chats are used to train models, and describes no opt-out.

On sharing, the policy says Janitor shares information with service providers for hosting, analytics, payment processing, customer support and marketing, and names Resend for service emails and Stripe for payments. It says it doesn’t sell personal data as defined under California law. The Safety page makes a broader promise, saying the company never sells your personal data or shares private conversations with third parties for advertising or commercial purposes.

The content license

The Terms of Service say you keep ownership of what you submit, and they also give Janitor a worldwide, non-exclusive, royalty-free, sublicensable and transferable license to use, reproduce, modify, publish, distribute and display it in connection with operating and improving the platform. The Terms define user content as any text, characters, images, audio, video or other material you submit. The license continues after you stop using the platform, but only as far as needed to maintain the platform and preserve content that was shared with or built into other users’ work.

Retention and sale of the company

The policy keeps personal data “only for as long as necessary” and lists the criteria, which include the length of your account, legal obligations, disputes and security. It states no number of days. The policy also has a business transfers section saying that in a merger, acquisition, reorganization, sale of assets or bankruptcy your information may be transferred to a successor entity, and that the company will notify you of the change and any choices you have. We found no sale or restructuring announced by Janitor AI.

A third party, VerifyWise, rates the disclosure in Janitor’s public documents at D, 40 out of 100, in an assessment dated October 13, 2026. The score reflects how much the documents disclose, and its snapshot lists the policy as last updated December 11, 2025, so its item-level findings may describe an older version than the one now on the site.

Janitor AI delete account

The privacy policy and the help center agree on the menu path. Open Settings, then the Security tab, scroll to the bottom and choose Delete account, then follow the prompts and enter the 6-digit one-time code the company emails to the address on the account. The help article, updated May 16, 2026, says support can’t delete an account for you and can’t help if you no longer have access to that email, and that the deletion removes your posts, characters, personas and profile information.

The privacy policy, updated later, adds detail. It says the account is deactivated first and the confirmation shows a scheduled deletion date. During the grace period you can sign back in and restore the account, and after permanent deletion the account can’t be restored. It also says deleting an account doesn’t immediately erase every record. Your email address and authentication-provider metadata may stay in deleted-account records for deletion notices, purchase or receipt recovery, security, fraud prevention, legal obligations or disputes, and you can email support@janitorai.com to ask for retained personal data to be deleted.

Two points are not covered in the pages we read. The help article doesn’t mention chat history by name, and neither document says how long the grace period lasts. The policy lists a right to a copy of your data in a structured, machine-readable format “depending on your jurisdiction,” requested by email, and the pages we read mention no export button. If you want a copy of your chats before you delete, ask support for it first, because deletion also cancels unused paid credits without a refund under the Terms.

Moderation, reporting and human review

Janitor’s Terms say it uses a combination of automated tools and human review. Automated systems scan content and may flag, restrict or remove it, and human reviewers assess content that users report. The Safety page adds that automated systems screen every piece of content before it is published and that full-time human moderators review reports around the clock. It gives these company-stated figures: more than 100,000 items reviewed by AI each day, more than 1 million automated reviews a week and more than 20,000 human moderation actions a month. They are the company’s numbers, and we found no outside check of them.

You can report a character with the Report button on its page, report a single chat message with the flag icon, open a ticket in the help center, or email support@janitorai.com, and the Terms say you don’t need an account to report illegal content. The Terms say the company aims to acknowledge reports within 7 days. Its Complaints Policy gives a faster target: acknowledgement within two business days and a review typically within seven business days, sooner for urgent safety matters. The company says it may confirm that action was taken without giving details.

Appeals run through the same support channels. A help article on bans says a rejected appeal is final and later tickets are closed without a reply. The content rules also say moderation decisions are final, that content can be removed without warning, and that moderators don’t have to explain their reasoning.

Self-harm protocol and crisis resources

Janitor publishes a Suicide and Self-Harm Protocol. It says responses on the platform are generated by AI systems, can make mistakes and aren’t a substitute for professional help. When a user expresses suicidal ideation, for example through certain phrases, the platform shows a crisis referral notification and restricts AI-generated responses about self-harm or suicide, and it logs the number of notifications for compliance reporting. The protocol page lists US 988 resources. The Safety page also lists Samaritans on 116 123 in the UK and Lifeline on 13 11 14 in Australia.

The content rules separate fiction from encouragement. Fiction can include suicidal themes under limits, and content that presents suicide as positive or inevitable, gives methods, or encourages self-harm is banned. If you are struggling, you can call or text 988 in the US, or contact your local helpline elsewhere.

Public chats, proxies and other ways your chats can travel

In March 2026 Janitor added public chats in open beta, which let you publish a chat as a snapshot that other people can read, react to, comment on and fork into their own chat. The announcement describes publishing as something you do, so a chat becomes public when you publish it. Chats are checked at that point, and unpublishing is one click. The announcement doesn’t say what happens to forks others already made. Before you publish, look at the persona name and any real details that appear in the text.

Janitor also lets users connect outside AI models. The Terms say that configuring an API proxy in your settings is allowed, and the help center describes a janitor+ Router that lets subscribers use supported third-party models inside Janitor. The Router article says nothing about which provider receives your chat. If you use a proxy or a third-party model, the company running that model has its own terms and retention, and you would need to read those separately. The Terms also forbid unofficial clients that sign in to your account, and say that using one can lead to the account being deleted.

The iPhone app and lookalike apps

The US App Store listing for Janitor: Interactive Stories, from the seller JanitorAI, INC, showed free with in-app purchases, an 18+ age rating with in-app controls, and 4.5 out of 5 from about 4,800 ratings on October 16, 2026. The listing’s privacy label, which the developer fills in and Apple doesn’t verify, lists contact info, user content, identifiers, usage data and diagnostics under data not linked to you.

Janitor’s February 2026 newsroom post warns that fake apps use its name and says the official app is published by Janitor AI Inc. A search of the US App Store for “janitor ai” returns other apps with similar names from other companies, so check the publisher before you install and never enter payment details in an unofficial app. If you installed a lookalike and signed in, the company advises deleting it and changing your Janitor password.

Subscriptions and refunds

Janitor has a free tier and a janitor+ subscription. The help center says the free tier has unlimited chats with a context limit of about 9,000 tokens, and that janitor+ has about five times that, which matters because the memory help article says older text falls out of a chat once the limit is reached and the character can forget earlier events. Stripe handles payments, and the company says it doesn’t see or store card numbers. Subscriptions renew automatically, you cancel in Settings under Billing or through the App Store for app purchases, and the Terms say there are no partial refunds for unused time except where the law requires it. Payment doesn’t change moderation: the Terms say a moderated or suspended account isn’t entitled to a refund.

What the sources leave open

Several questions have no answer in the documents we read. The policy says nothing definite about whether chats train models, the retention period is undefined, and the grace period before permanent deletion isn’t given in days. We found nothing on whether the creator of a character can see chats with it, nothing on who can read a chat beyond human review of reported content and support or security access to account information, and nothing in the privacy policy on which model providers handle the text of chats. The Safety page and the help article list different countries for age verification. The help center and the privacy policy describe deletion with different levels of detail, and the policy is the newer document.

Questions about any of these can go to support@janitorai.com.

Checklist before you use Janitor AI

How Riho compares

Janitor AI is a platform of user-made characters, and Riho is one girlfriend. Riho is made for adults, runs on iPhone and hasn’t launched yet, and in the app you can read what she remembers about you. Her privacy page covers what the website forms collect and how the iPhone app handles your data. The questions in the checklist above apply to every companion or roleplay app, including ours, and the safety checklist for AI girlfriends walks through them in order. Our guide to why Riho is one girlfriend with no catalog explains the difference in design, and the best AI girlfriend apps list places several apps side by side.

Early access to Riho opens from the button below.

Frequently asked questions

Is Janitor AI safe to use?

It is an adults-only platform with published terms, a privacy policy dated September 17, 2026, human review of reported content and a self-harm protocol. It sets no retention period, doesn’t say whether chats train models, and keeps some account records after deletion. Use it as an adult, keep identifying details out of chats, and read the policy before you pay.

Is Janitor AI safe for kids?

No. Janitor’s Terms say it is intended solely for users who are at least 18, the iPhone app is rated 18+, and the content rules allow adult sexual content between consenting adult characters. Users the company suspects are under 18 are banned.

Does Janitor AI require age verification?

Only in some places. A help article says users connecting from Australia, Brazil or the UK complete a one-time check through k-ID, and that users elsewhere see no change. Everywhere else the 18+ rule is a term you agree to.

How do I delete my Janitor AI account?

Open Settings, then Security, scroll to the bottom and choose Delete account, then enter the 6-digit code emailed to you. The privacy policy says the account is deactivated first and can be restored during a grace period, and that some records may remain afterwards.

Can people see my Janitor AI chats?

The public chats feature turns a chat into a snapshot others can read and fork once you publish it, and unpublishing is one click. The documents we read say human reviewers assess content that is reported to the company, and that support and security staff may access relevant account information.