Is Janitor AI Safe? What Happens to Your Chats

Janitor AI is a legitimate, widely used character-chat platform with public content rules, an adults-only policy and self-service account deletion. It is not a private diary. Your chats are stored on Janitor's servers, and if you connect an outside model, your messages also go to that provider. How safe it is for you depends mostly on two choices: what personal information you put into chats, and which model or proxy you connect.

This guide is based on Janitor AI's official help center, checked on October 23, 2026, and on its Terms of Service (updated October 8, 2026), Privacy Policy (updated September 17, 2026) and Content Guidelines. janitorai.com blocked our direct requests, so the legal pages were read from Internet Archive copies made October 13 to 19, 2026. We did not test a paid account.

Assume yes, technically. Chats are tied to your account and stored in Janitor's database: the help center's missing chats article refers to chats lost "during server migrations" and older chats waiting for "a database migration" back into the "main database." Anything a service stores on its servers can, in principle, be accessed by the operator.

The Privacy Policy lists "chat conversations, messages" among the content Janitor collects and says data is used for "research and development to improve our AI systems". It does not say plainly whether chats train JLLM (Janitor's own model), and it sets no fixed retention period: data is kept "only for as long as necessary". The Terms say Janitor uses "a combination of automated tools and human review" to detect illegal content. Janitor says it does not sell personal data under the CCPA. Treat every chat as readable by the company.

Beyond automated scanning, moderation relies on reports. The help center says reports of bots and comments are reviewed by the moderation team, and that Janitor does not discuss individual moderation actions in public, citing user privacy and GDPR.

The official help pages do not describe any feature that shows your private chats to other users, and none of the articles we read mentions chat sharing. What other users can see is public content: bots you publish, comments, your profile and its custom CSS.

Two cases deserve care:

• Bot creators. The help center does not say whether the author of a bot can see conversations with it. We found no statement either way, so do not assume either.
• Screenshots. Anything you paste into a public Discord, subreddit or comment is public, whatever the platform's settings.

Janitor's default model, JLLM, runs on Janitor's side. Users can also connect outside models in two ways: the paid janitor+ Router, which handles model access inside Janitor without your own key, or their own API settings, where they enter a provider endpoint and an API key. Janitor's glossary calls this a proxy: "the middleman between your chat and the model provider."

The safety trade-offs:

• More companies see your messages. With an outside model, each message in that chat is sent to the provider (for example, a model marketplace) and handled under its privacy policy and usage rules, as well as Janitor's.
• Your key is a password. Janitor's own proxy guide tells users to store the key "like a password." Someone holding it can spend your credit. Set a spending cap at the provider, keep the key out of screenshots and bot descriptions, and revoke it when you stop using it.
• Unknown proxies can log everything. Proxy links shared by strangers route your chats and your key through a server you cannot audit. If you would not hand that person your messages and password, do not use their endpoint.
• Filters and rules do not disappear. Connecting a different model does not change Janitor's content guidelines or the provider's own policies. This guide does not cover ways around either.

If you want hosted models with one bill and no keys, the Router or a service that hosts its own models is simpler. See Janitor AI alternatives.

The Terms say the platform is "intended solely for users who are at least 18 years of age" and ban any content that "contains any depiction of individuals under 18 years of age". The help center and Content Guidelines add that accounts reasonably suspected to belong to minors are permanently banned. The minor-related content policy is zero-tolerance: characters or personas coded as under 18 are removed regardless of context.

Since 2026, users connecting from Australia, Brazil and the UK must complete a one-time age verification through k-ID, by on-device facial age estimation, an ID scan or an existing AgeKey. According to the article, k-ID does not keep the document or face data after the check, and Janitor receives only a yes or no on whether you are over 18. Detection is by IP address only.

Characters with sexual or romantic themes must be tagged Limitless, and the official mobile app starts with NSFW content disabled to meet app store rules. The Content Guidelines also ban "obvious or recognizable depictions of real, non-celebrity individuals", and say creators, influencers and adult performers do not count as celebrities.

Card payments for janitor+ on the web are processed by Stripe; the subscription FAQ says Janitor does not see or store card details. App subscriptions are billed and cancelled through the App Store or Google Play. janitor+ costs $12.99 a month or $99 a year (checked October 24, 2026). The Terms say subscriptions renew automatically at the end of each billing period unless cancelled, and that prices may change. Subscriptions are generally non-refundable unless the terms or law require otherwise, Router credits are non-refundable except where law requires, and deleting the account cancels unused credits without refund. Billing questions go to [email protected].

The help center says the iOS and Android apps it links are "the only official JanitorAI apps." Unofficial apps or sites using the Janitor name can collect logins or keys; install only from the official app page.

1. Keep identifying details out of chats and personas: full name, address, workplace, school, phone number, photos of real people.
2. Do not write about real, identifiable people in roleplay.
3. Prefer JLLM or the janitor+ Router over proxies from unknown sources.
4. If you use your own API key, set a spending limit and rotate the key regularly.
5. Use a unique password and an email you control; account deletion requires a code sent to that email.
6. Read the current Terms and Privacy Policy before sharing anything sensitive; both were updated in September and October 2026.

For a broader checklist that applies to any companion app, see our AI companion privacy guide.

Per the help center: open Settings, go to the Security tab, scroll to the bottom and click Delete account, then enter the 6-digit code sent to your email. Deletion removes posts, characters, personas and profile information. The Privacy Policy adds that the account is deactivated first and can be restored during a grace period; after permanent deletion some records, including your email address, may be kept for billing, security and legal reasons. The help center directs users to this in-app flow, and without access to the account email the deletion cannot be completed. If you used your own API key, revoke it at the provider as well, because deleting the Janitor account does not cancel the key.

• Missing chats
• janitor+ subscription FAQ
• janitor+ Router
• Proxy quickstart
• JLLM glossary
• Age verification
• Account deletion
• Minor-related content
• Reporting content
• Moderation transparency
• Official apps
• Terms of Service
• Privacy Policy
• Content Guidelines

Help center checked October 23, 2026. Terms, Privacy Policy and Content Guidelines read from Internet Archive copies made October 13 to 19, 2026, because janitorai.com blocked direct requests.