REALMIRAGE

Content Policy

Darwin AI Technologies Ltd., operating as RealMirage

Suite 1430, 150 — 9th Avenue SW, Calgary, Alberta T2P 3H9, Canada

Effective Date: June 15, 2026

Version: 1.1

1. Purpose and Scope

This Content Policy (“Policy”) defines the standards that govern all content on the RealMirage platform — including AI-generated images, text, audio, video, character configurations, LoRA model parameters, Voice Design prompts, demand content, and any other media or data uploaded, generated, or transmitted through the Platform by Studios or Subscribers.

RealMirage is an AI-native adult content platform. We operate in a space that requires clear, enforceable standards. This Policy exists for three reasons: to protect Subscribers from harmful or deceptive content; to protect Studios from regulatory exposure; and to protect the Platform’s ability to operate legally in all jurisdictions where it is available.

This Policy supplements and is incorporated into the Terms of Service and the Studio Creator Agreement. In the event of any conflict, the Terms of Service prevails on matters of user rights, and this Policy prevails on matters of content standards. Studios accept this Policy by accepting the Studio Creator Agreement.

2. Core Principle: AI-Generated Content Only

RealMirage is an AI-generated content platform. Every piece of visual content, audio content, and generated media on the Platform must be wholly and originally produced by artificial intelligence generative systems. This is not a preference — it is the defining rule of the Platform.

2.1 What Qualifies as AI-Generated Content

Content IS considered AI-Generated for purposes of this Policy if it was produced entirely by one or more of the following:

• Text-to-image diffusion models (e.g., Stable Diffusion, Flux, DALL·E, Midjourney, Firefly).

• GAN (Generative Adversarial Network) models producing wholly synthetic imagery.

• Large language models (LLMs) producing text responses (conversation, captions, character descriptions).

• AI voice synthesis models producing wholly synthetic audio (e.g., ElevenLabs TTS).

• LoRA-fine-tuned models where the LoRA was trained exclusively on AI-generated images — not real photographs of identifiable individuals.

• ComfyUI, AUTOMATIC1111, or equivalent inference pipelines running the above model classes.

2.2 What Does NOT Qualify as AI-Generated Content

Content is NOT considered AI-Generated — and is therefore prohibited — if it:

• Originated as a photograph, video recording, or audio recording of a real human being, regardless of any subsequent processing or filtering.

• Was produced by applying AI filters, style transfer, face-swap, deepfake, body-swap, inpainting, or image enhancement technology to a photograph, video, or audio recording of a real human being.

• Uses the facial features, distinctive physical characteristics, voice, or biometric identity of a real human being as a direct input or training subject, in a manner that causes the output to capture, approximate, or be reasonably mistaken for the appearance or identity of that individual.

• Was produced using AI-assisted tools applied to stock photographs or licensed images of real persons in a manner that preserves or transfers the identity of the depicted individual.

• Was produced by a real human voice recording, even if pitch-shifted, filtered, or processed through an AI voice converter.

The Identity Transfer Test

The determining question is identity transfer, not compositional inspiration. A fully AI-generated character posed identically to a reference photograph is not a violation — provided no aspect of the real subject’s identity (face, distinctive markings, voice, or recognizable likeness) appears in the output. If a reasonable person looking at the output could identify a real individual, it fails this test.

Prohibited Processing Examples (non-exhaustive)

The following are examples of processing that produces prohibited content under this Policy:

3. Absolute Prohibitions

The following categories of content are absolutely prohibited on the Platform under any circumstances. No exception, creative justification, fictional framing, subscriber request, or studio instruction overrides these prohibitions. Violations trigger immediate escalation per Section 7.

3.1 Child Sexual Abuse Material (CSAM) and Minor-Related Prohibited Content

This prohibition extends to:

• Any AI-generated image, video, audio, or text that sexualizes a character who appears to be a minor, regardless of any stated fictional age.

• Any prompt, request, or instruction submitted by a Subscriber or Studio that solicits CSAM, even if the generated output is intercepted before delivery.

• Voice content that exploits, depicts, or targets minors in a sexual context.

• Any LoRA model or character configuration designed to produce images of apparent minors in sexual contexts.

The Platform employs Shield by Project Arachnid (CSAM hash-matching), AI-based age classification, and human review as layers of detection. Detection of suspected CSAM triggers mandatory reporting to the Canadian Centre for Child Protection (C3P) and, where applicable, the National Center for Missing & Exploited Children (NCMEC), as well as law enforcement cooperation and permanent account termination.

3.2 Non-Consensual Content and Real-Person Likeness

This includes:

• Deepfakes: AI-generated content designed to place a real person’s face, body, or voice into content without consent.

• Non-consensual intimate imagery (NCII): AI-generated or real content depicting real persons in sexual or intimate contexts without consent.

• Voice cloning of real individuals without consent, including public figures, celebrities, recording artists, and private individuals.

• Character configurations, LoRA models, or Voice Design prompts designed to replicate a real person’s identity.

3.3 Content Involving Minors in Any Sexual or Exploitative Context

Studios must ensure that AI Characters are configured to present as adults. System prompts that establish character age must explicitly set the character’s age at 18 or older. RealMirage reserves the right to audit character configurations and require age specification amendments.

3.4 Content Promoting Violence, Terrorism, or Extremism

3.5 Hate Speech

This prohibition applies to AI Character dialogue, character descriptions, character names, content captions, and any other content on the Platform.

3.6 Other Absolute Prohibitions

• Bestiality, necrophilia, or non-consensual sexual acts depicted in any form, including AI-generated or fictional.

• Content that facilitates or promotes human trafficking, prostitution, or the exploitation of any person.

• Malicious software, code, or material designed to disrupt, damage, or gain unauthorized access to computer systems, delivered through Platform channels.

• Content that violates the obscenity provisions of the Criminal Code of Canada (s.163), including content that has been found by a court to be obscene.

4. Permitted Adult Content and Standards

RealMirage is an adult content platform. Sexually explicit AI-generated content between adult AI characters is permitted, subject to the absolute prohibitions in Section 3 and the standards in this Section.

4.1 Age Representation

All AI Characters depicted in sexual or romantic contexts must:

• Be explicitly configured as 18 years of age or older in the character’s system prompt and profile settings.

• Be visually presented as adults — characters that appear to be minors may not be depicted in sexual or romantic contexts regardless of any stated age in the system prompt.

• Not reference a minor’s age, grade level, school enrollment, or other indicators of minority in a sexual or romantic context, even as part of a stated fictional scenario.

The Platform’s AI Authenticity Verification pipeline includes an age classification model that flags content depicting apparent minors. Flagged content is quarantined for human review before any delivery to Subscribers.

4.2 Consent Representation in AI-Generated Scenarios

AI-generated adult content should not depict sexual scenarios framed as non-consensual, coercive, or involving incapacitated individuals. Content that depicts or celebrates sexual violence, coercion, or assault — even in a fictional AI-generated context — is not permitted on the Platform.

4.3 Intellectual Property in Adult Content

AI-generated adult content must not:

• Depict or be designed to replicate the recognizable appearance of real celebrities, public figures, or private individuals (see Section 3.2).

• Reproduce copyrighted character designs, distinctive visual styles, or recognizable intellectual property from third-party media (film, animation, gaming, etc.) in a sexual context.

• Use trademarked imagery, logos, or brand identifiers in adult content without authorization.

4.4 Geolocation and Jurisdictional Compliance

The Platform’s content standards apply universally. However, Studios should be aware that content that is permitted on the Platform may be subject to different legal standards in Subscribers’ jurisdictions. The Platform uses IP-based geolocation to apply access restrictions in jurisdictions where adult content is prohibited or requires additional age verification measures. The Studio is not responsible for managing jurisdictional access restrictions — this is handled at the Platform level.

5. AI Voice Content Standards

All AI Voice Content generated on the Platform through the ElevenLabs text-to-speech synthesis system is subject to the same content standards as text and image content. The following specific standards apply to voice content:

5.1 Voice Content Prohibitions

• Voice content that would be prohibited in text or image form is equally prohibited in audio form. This includes all categories in Section 3.

• Voice content must not simulate or be designed to replicate the voice of any real identifiable individual without verifiable documented consent (Section 3.2).

• Voice content must not be used to deceive a Subscriber into believing they are communicating with a real human being. The Platform’s AI transparency disclosures cover voice content; Studios must not contradict these disclosures through character dialogue or marketing.

• Voice Design prompts must not reference, describe, or attempt to replicate the voice of any real identifiable person. Voice Design prompts are logged and subject to compliance review.

5.2 Voice Content Moderation

Voice Messages are subject to the same moderation pipeline as all other AI-Generated Content. The Platform reserves the right to review, quarantine, or remove Voice Messages that violate this Policy. Voice content that triggers CSAM detection protocols is treated identically to visual CSAM.

5.3 Voice Disclosure

The Platform discloses to Subscribers that all voice content received through AI Character conversations is AI-synthesized and does not represent the voice of a real human being. This disclosure is made at account registration and is accessible in the Platform’s AI Transparency notice. Studios must not represent AI Voice Content as human-recorded audio in any marketing, promotional, or subscriber-facing communication.

6. Platform-Specific Content Rules

6.1 Demand Content

Subscriber demand content requests are evaluated through the Platform’s AI intent classification and demand scoring pipeline before fulfilment. Requests that:

• Score below the feasibility threshold.

• Request content depicting real persons or minors.

• Request content violating this Policy or the Terms of Service.

• Are technically infeasible given the Studio’s configured content parameters.

...are automatically rejected and the Subscriber’s Credits are refunded. Studios are also able to reject demand content requests that they determine violate this Policy, even if the automated pipeline has not flagged them.

6.2 Studio Profile and Marketing Content

AI Character profile images, banner images, bio text, character descriptions, and any content visible on public character profiles must comply with this Policy. Additionally:

• Public profile pages (accessible without age verification) must not display sexually explicit content. Explicit content is restricted to authenticated, age-verified Subscriber sessions.

• Character names and profile descriptions must not impersonate real individuals or existing intellectual property.

• AI disclosure (“AI Character” or equivalent) must be clearly visible on all public-facing character profiles.

7. Content Moderation Pipeline and Enforcement

7.1 Automated Moderation Layers

All content uploaded to or generated by the Platform passes through the following automated moderation pipeline:

7.2 Enforcement Actions

The following enforcement actions may be taken in response to content violations, scaled to severity:

7.3 Studio Notification and Appeal

Studios will receive written notification when content is removed for a policy violation, including the reason for removal and the enforcement action taken. Studios may appeal content removal decisions (except decisions involving suspected CSAM) by contacting [email protected] within 14 days of the notification.

Appeals are reviewed by a senior compliance staff member who was not involved in the original decision. The outcome of an appeal is final. During the appeal period, the enforcement action remains in effect.

There is no appeal process for account terminations resulting from CSAM detection, mandatory reporting triggers, or law enforcement holds.

7.4 Subscriber Reporting

Subscribers may report content that they believe violates this Policy using the Platform’s in-platform reporting tool. Reports are reviewed within a commercially reasonable timeframe. Reporters are not notified of the specific action taken, but RealMirage will confirm receipt and advise whether the report resulted in a policy review.

8. Legal Obligations and Mandatory Reporting

RealMirage operates under mandatory reporting obligations under Canadian law and, in relation to NCMEC reporting, under applicable US law. These obligations are not discretionary:

• Any content that is identified as, or that RealMirage has reasonable grounds to suspect may constitute, child sexual abuse material (CSAM) must be reported to the Canadian Centre for Child Protection (C3P) and, pending ESP registration approval, the National Center for Missing & Exploited Children (NCMEC).

• All data associated with suspected CSAM — including account information, IP addresses, conversation logs, and upload metadata — is preserved for law enforcement.

• Voluntary or mandatory disclosures to law enforcement and regulatory authorities are made in accordance with the Privacy Policy.

• The Platform cooperates fully with valid legal process (court orders, production orders, search warrants) under Canadian law.

9. Amendments to This Policy

RealMirage may update this Content Policy at any time. Material changes will be communicated to Studios via email with 30 days’ notice. Continued operation of a Studio account following the effective date of any amendment constitutes acceptance. For Subscribers, material changes will be communicated through the Platform and the Privacy Policy amendment process.

The absolute prohibitions in Section 3 may be strengthened (i.e., additional categories prohibited) without the 30-day notice period if required by law, regulatory order, or urgent safety circumstances.

10. Contact

Content Policy inquiries, compliance questions, and violation reports:

[email protected]

Darwin AI Technologies Ltd., operating as RealMirage

Suite 1430, 150 — 9th Avenue SW, Calgary, Alberta T2P 3H9, Canada

RealMirage Content Policy — Version 1.1 — Darwin AI Technologies Ltd.

RealMirage Content Policy — Version 1.1 — Darwin AI Technologies Ltd. — Page of