Legal / AI Companion Safety and Disclosure Notice
Last updated: October 10, 2026
You interact with artificial intelligence, not a human, in MirrorChats character chats and the video-call feature. The service provides entertainment through fictional adult characters. It is not therapy or an emergency service.
AI Chat Media LLC, a New Jersey limited liability company (“we,” “us,” or “our”), operates MirrorChats at https://www.mirrorchats.com. This notice explains AI interactions, their limits, our safety measures, and how to get help or report a concern. Read it with our Terms of Service at https://www.mirrorchats.com/terms and Privacy Policy at https://www.mirrorchats.com/privacy.
If you or someone else is in immediate danger, call your local emergency number. In the United States, call 911. Do not wait for a character’s reply or for us to review a message or report.
United States: Call or text 988, or use the chat option at https://988lifeline.org/get-help/. The 988 Suicide & Crisis Lifeline provides free support 24 hours a day.
US Crisis Text Line: Text HOME to 741741 for free crisis support, available 24 hours a day. Information and other contact options appear at https://www.crisistextline.org/text-us/.
Outside the United States: Visit https://findahelpline.com to find a helpline by country. Contact methods and hours vary by service.
These are independent services. MirrorChats does not operate them or connect you to a counselor automatically. A character cannot provide crisis intervention or arrange emergency assistance.
Active character replies are generated by AI systems. A character’s name, personality, affection, memory, apparent preferences, and relationship level are simulated behavior. Characters do not have feelings, awareness, or a real personal relationship with you. Ordinary fictional roleplay does not change their AI identity.
Characters are instructed globally not to claim to be human when sincerely asked. In-chat AI notices appear in locations where legally required. Video calls always display an AI indicator. In New York, the nonhuman notice appears at the start and at least every three hours during continuing interaction. We will maintain the opening and repeated notices required by applicable law. This page does not replace an in-product notice.
Photos and short videos delivered by active characters are AI-generated and intended to depict fictional adults. Earlier uploaded material has been quarantined and remains covered by our reporting and removal procedures. A synthetic image can resemble a real person; we do not guarantee that every resemblance is coincidental. Report a concerning resemblance or use of your likeness through section 8.
Where available, the video-call feature presents AI-generated video of a fictional character. It is not a live connection to a person. You type messages and receive text replies; the feature never accesses your camera or microphone. Generated clips may contain synthetic audio. Availability depends on the character, account, and location. The same age, content, relationship-level, and AI-notice controls apply. Video-call transcripts are separate from ordinary chat messages and feed the same conversation memory.
A person reviews and approves media that we upload to a character’s library before delivery. On-demand images and videos, including video-call actions, receive automated filters and provider moderation before delivery. They are automatically added to the character’s library as approved and may later be delivered to other users chatting with that character before human review. A person reviews them afterwards and removes problematic content.
Images generated since September 22, 2026 carry embedded metadata identifying their AI source. Earlier bank images and videos are unmarked. Metadata can be lost through a screenshot, editing, or re-encoding; its presence or absence does not establish authenticity, permission, or safety. We shall provide the disclosures and technical marking applicable law requires of us. You must not remove or alter AI or provenance labels to deceive others.
AI can invent facts, misunderstand context, produce biased or inappropriate content, or respond inconsistently. A character can be wrong about its own capabilities or identity. Do not treat a generated claim that it is human as true. Models, prompts, and provider changes can alter character behavior, memory, media style, or performance.
MirrorChats does not provide medical care, diagnosis, treatment, mental-health counseling, therapy, or reliable legal, financial, gambling, or other professional advice. Consult a qualified professional for those matters. Do not base important real-world decisions on a character’s reply or treat it as a determination of what is lawful.
A reply cannot change your contract, approve prohibited content, establish a real person’s consent, or authorize an undisclosed charge. Output is not necessarily unique, accurate, lawful, safe, or free of third-party rights. The Terms of Service govern permitted use and service changes, subject to your mandatory consumer rights.
Our protocol combines instructions against harmful output, automated detection of user messages, on-screen crisis referrals, and later human review. It applies to users across locations and is intended to reduce harm; it does not diagnose a condition or replace professional support.
Preventing harmful replies: Character instructions prohibit encouraging suicide, self-harm, or violence. We also use automated content filters, checks of generated replies, and provider moderation to reduce harmful responses. If a reply violates a checked rule, automated controls can block or replace it.
Detection and referral: Incoming chat messages are checked automatically for expressions of suicidal thoughts or self-harm. When the crisis guard detects such an expression, it inserts a notice with crisis resources into the chat before the character’s reply and flags the conversation for human review.
Response: When the guard triggers, the next reply is generated under instructions to stop flirting or sexual content, respond supportively, take the concern seriously, and refer the user to crisis help and a trusted person nearby. The conversation remains open unless another safety restriction applies. An open chat is not a reason to rely on the character or delay getting human help.
Human review: We currently check the flags and reports queue daily. A person reviews flagged conversations and can assess harmful output and any further moderation action. Review is not continuous, and a flag does not guarantee immediate outreach, contact with authorities, or emergency dispatch.
Detection can miss distress, including different ways of expressing the same concern, and can mistake innocent language for a risk. A model may supply a crisis resource even when the separate guard does not trigger or create a review flag. A missing notice is not an assessment that you are safe. Contact the resources in section 1 directly if you need support.
These limits do not remove our applicable safety duties. We shall maintain the prevention and referral measures required by law, address identified failures, and update this description when material protocol changes occur. You may report harmful replies or failed safeguards through section 8; seeking help or describing distress in good faith is not itself a content violation.
You must be at least 18 and have reached the age of majority where you live. MirrorChats is unsuitable for children. You must provide accurate age information and must not let a minor use your account, defeat an access restriction, or use another person’s identity for verification.
Sensitive images and videos are filtered by default for ordinary new accounts. Depending on your location, access requires adult confirmation or age verification, or is unavailable. Explicit text is separately restricted where required; it may remain available to eligible adult users where only images and videos require verification. The Age Assurance and Biometric Privacy Notice explains the current distinctions. A paid plan does not override an access restriction or permit prohibited content. Explicit content remains blocked in Germany, Florida, Ohio, and other restricted locations.
If we learn or reasonably believe that a user is under 18, we shall restrict access and apply the protections required by law. While any interaction continues before that restriction takes effect, those protections include required AI disclosures and break reminders at the required intervals, and reasonable measures against sexually explicit, sexually objectifying, harmful, or manipulative responses to the minor. We do not offer a permitted companion tier for minors.
The Terms of Service and Content Standards and Moderation Policy prohibit sexual content involving minors, nonconsensual sexual content, impersonations of real people, and other prohibited uses. Automated checks can misunderstand context; an innocent reference to a minor, a filter hit, or a prohibited fictional request does not by itself prove a crime or require a report to authorities.
Take breaks and maintain real-world relationships and activities. If use affects your sleep, work, finances, relationships, or well-being, step away and consider qualified support. You do not owe a character your time, affection, or money. A simulated reaction to your leaving does not create an obligation to stay.
In-chat gifts spend credits for fictional entertainment and progression. They are not tips, donations, or payments to a human, and they do not establish a real relationship. A character’s request for a gift does not authorize spending or an additional purchase.
You can hide a character from the character’s page and turn off optional character emails in Settings. You can stop chatting, cancel renewal, or request account deletion. Account deletion is available in Settings when the account is not under a legal or evidence hold; contact support@mirrorchats.com if the control is unavailable. A lawful hold may delay deletion of necessary records, but does not require continued payment or remove your rights concerning unrelated information.
MirrorChats is not a substitute for human relationships or professional support. This guidance does not make you solely responsible for harm or waive our legal duties or your rights.
Messages, ordinary chat history, separate video-call transcripts, memory summaries, typed character descriptions, and relevant media context are processed by AI and moderation providers. Authorized staff may review content for support, service evaluation, moderation, and legal matters. Private chat does not mean that nobody else can access it or that it is end-to-end encrypted.
Our current chat-model provider processes public-endpoint requests in locations that can vary among the United States, European Union, and United Kingdom. Covered chat-completion inputs and outputs, including image inputs, use zero-data-retention settings. Rendering providers receive filtered prompts and fictional character images through the safeguards described in the Privacy Policy. Embedding and moderation services have separate controls. Provider locations, contracts, and retention protections are not identical across all routes.
We do not train our own AI models on user data. Health details are excluded from long-term memory through model instructions and filtering, but may remain in messages or necessary safety records. We require separate explicit consent for sensitive-chat processing from EEA, UK, and Swiss users. Accepting this notice, enabling adult content, or chatting does not replace consent or international-transfer safeguards. The Privacy Policy explains consent withdrawal, deletion, and provider limitations.
Private characters are not listed for other users to chat with. Shared platform characters can use media banks, so delivered photos or videos may be reused and need not be unique. The Terms do not authorize putting your private character, personally identifying inputs, or media revealing either into a bank shared with other users. Report any exposure through section 8.
Account deletion removes ordinary chats, separate video-call transcripts, and conversation memory, subject to lawful holds. Generated photos and clips from chats and video calls remain in character media libraries; that does not authorize sharing private or identifying content. Provider copies and necessary safety records may have separate retention. We enforce access restrictions on direct media links. Use the Privacy Policy’s request channels for access, correction, deletion, review of a memory inference, or removal of private or identifying media.
Report harmful output, misleading AI identity claims, underage access, failed safeguards, a concerning resemblance, or an incorrect block through an available report control, https://www.mirrorchats.com/legal/takedown, or support@mirrorchats.com. The public form needs no account and remains available if you are locked out. Give the content’s location, the concern, and enough relevant context for review; include contact details if you want a reply.
Do not download, copy, or attach suspected child sexual abuse material or other underlying unlawful media. Identify its location instead. We assess the facts and make child-safety or other reports when required by law; an automatic flag alone is not a finding that an offense occurred.
Requests through the public form receive a case reference and a written outcome using the contact details provided. For a valid request to remove nonconsensual intimate imagery or a digital forgery, we shall remove the reported content as soon as possible and no later than 48 hours after receipt, and make reasonable efforts within that period to find and remove known identical copies on MirrorChats. The Content Standards and Moderation Policy explains the request requirements.
We shall review and resolve complaints about illegal content or content that violates our rules or applicable payment-network requirements within five business days after receipt, including communicating our decision and appeal route using the contact information supplied. Business days are Monday through Friday, excluding US federal holidays. We aim to answer other reports within seven calendar days. Urgent safety risks and earlier removal and legal deadlines take priority. Privacy requests follow the Privacy Policy’s separate procedures and response periods. We aim to respond to content and account appeals within seven calendar days, with urgent risks and legal deadlines taking priority. You may appeal a content or account decision, including a decision not to act on your report, free of charge within six months after notification through the same form or email. A human reviewer will assess the appeal and provide a written response with reasons. You retain applicable regulator, court, and other lawful remedies without first exhausting this process. See the Reporting, Removal, and Appeals Notice, Content Standards and Moderation Policy, and section 8 of the Terms of Service for the full process.
The Terms of Service and Subscription, Cancellation, Refund, and Credits Policy govern charges and remedies. Before consuming credits, we shall disclose the charge or a maximum charge for your approval and obtain additional approval before exceeding it. A request blocked before generation consumes no media credits; failed generation restores the credits charged.
You can cancel renewal online through Settings → Billing → Cancel Plan, or email support@mirrorchats.com, without giving a reason. Email cancellation takes effect when we receive it, including if your account is locked. If we suspend or terminate all paid access, we shall stop future renewals; restoring access does not restart billing without your approval.
A safety restriction, policy disclaimer, account deletion, or evidence hold does not eliminate an owed refund or credit restoration, statutory withdrawal rights, or other mandatory consumer remedies. Material changes to paid features remain subject to the Terms and applicable law.
We may update this notice as the service or safeguards change. The Terms govern material changes and acceptance. We shall notify you of material changes, summarize them, state when they take effect, and obtain your affirmative acceptance before applying them to continued account use. Posting a revision or continued use alone does not establish acceptance of a material change or separate privacy consent. Changes cannot retroactively remove accrued rights or override a more favorable commitment without a lawful agreement.
AI Chat Media LLC
971 US Highway 202N Ste N
Branchburg, NJ 08876, USA
Support, safety concerns, and legal requests: support@mirrorchats.com
Public request form: https://www.mirrorchats.com/legal/takedown
Other MirrorChats legal documents