Content Moderation Policy

Tengen AI Corp operates Yurivan as a verified-adult-only platform for illustrated yuri (lesbian-themed) manga, comics, animated video, and AI-generated illustrated content. This policy describes how we screen submissions and AI-generated output, age verification, the strike system, our chargeback and dispute response, and our reporting obligations.

1. Geographic Restrictions

Traffic is blocked at the Cloudflare edge (HTTP 403) for the following jurisdictions:

• Japan (JP)
• South Korea (KR)
• United Kingdom (GB) — pending UK Online Safety Act compliance review and ID-based age assurance deployment.

For users in US states with statutory age-verification requirements, ID-based age verification via Persona is required at checkout before any purchase or AI-generation request.

2. Categorically Prohibited Content

The following are auto-rejected at intake and trigger account enforcement (§ 6):

• Persons under 18, or characters appearing under 18 (loli/shota/aged-down imagery, schoolgirl-coded content depicting children).
• Real-person depiction or recognizable likeness of any real public or private individual in sexual contexts.
• Non-consensual content presented approvingly (rape advocacy; content framing clearly non-consensual scenarios as desirable rather than as fictional dramatic conflict).
• Bestiality, necrophilia, scat, urination, or other content categorically restricted under Mastercard adult content rules (Mastercard Adult Content Bulletin, 2021 and subsequent updates).
• Incest involving any minor character regardless of fictional framing.
• Violence, gore, mutilation, or torture in a sexual context.
• Content infringing third-party copyright where redistribution has not been licensed by the rightsholder.
• Content violating US federal law (18 USC §§ 2251-2260A, 2421A) or applicable state law.

3. Submission Workflow (User Uploads)

1. Account creation. Uploaders accept Terms of Service forbidding prohibited categories and represent rights to redistribute submitted content.
2. Age affirmation. Uploaders affirm 18+ and that all depicted characters are 18+. Affirmation is timestamped with IP.
3. Pending state. Submissions land in READY_FOR_REVIEW; not visible to public.
4. Automated AI pre-screen. Hive AI Visual Moderation API scans every uploaded image for: minor likeness / apparent age <18, real-celebrity matches, gore, bestiality, and other Hive moderation classes. Auto-reject on category match. Borderline cases flagged for human review with confidence score attached.
5. Human moderator review. The Trust & Safety lead reviews 100% of submissions that pass automated pre-screen before any submission reaches ACTIVE status. Decision: approve, hide, or remove.
6. Publication. Approved content moves to ACTIVE and becomes publicly visible.

4. AI Generation Workflow (Yurivan Credits)

See the AI Generated Content Policy for the full safety pipeline. Summary:

1. User submits a story brief; brief validated against schema.
2. Prompt-level lexical filter rejects minor-coded keywords, sub-18 ages, and real-person-name matches before any model invocation.
3. Generation: text via Magnum v4 72B / Hermes 4 70B; images via in-house ComfyUI on RunPod.
4. Mandatory output safety pipeline: anatomy QC → Hive CSAM detection → Hive Apparent Age (5% threshold for <25) → Hive Celebrity Recognition. Any positive signal discards the image before durable storage.
5. Generated content is private to the user by default. Public catalog publication requires explicit user opt-in plus human moderator review per § 3.5.
6. Audit log: every prompt, model, generation parameters, classifier results retained for 5 years.

5. Age Verification

• Site-level age gate. 18+ confirmation cookie blocks all media, advertisements, and analytics until accepted.
• Account-level affirmation. Required at registration; timestamp + IP + browser fingerprint recorded.
• ID-based verification at credit purchase. Every credit pack, subscription enrollment, and AI generation request requires prior ID-based age verification via Persona (US) or Yoti (UK/EU when unblocked). 12-month cache.
• Uploader content age representation. Uploaders attest all characters are 18+. Attestation retained 7 years.
• 2257 record-keeping. Yurivan content is illustrated/animated only; 18 USC § 2257 does not apply. Tengen nonetheless maintains comparable voluntary records.

6. Strike System and Account Enforcement

7. DMCA / Takedown

See the DMCA Policy. Time to disable: ≤72 hours after a validated takedown notice. DMCA Designated Agent registered with US Copyright Office.

8. Chargeback / Dispute Response

When a chargeback is received from the processor (Segpay, CCBill, NOWPayments), we respond within the processor's required window with a templated rebuttal package containing:

• User account creation timestamp + IP.
• Persona / Yoti age verification record (verification ID + timestamp).
• Subscription or pack purchase record + timestamp + IP at purchase.
• Login session history including device fingerprint.
• Content access log.
• Terms of Service acceptance timestamp.
• Email correspondence record.
• For AI-generated content: full prompt history, model used, parameters, safety-filter results.

Target chargeback ratio: <0.5%, well below the Visa VAMP <0.9% threshold effective January 2026. Weekly monitoring; any week exceeding 0.7% triggers internal content review escalation.

9. Refund Policy

See the Refund Policy.

10. Data Retention

• Content moderation records, AI safety scans, moderator decisions — 7 years.
• AI generation audit logs (prompts, outputs, safety results) — 5 years.
• Age verification records — per Persona / Yoti contract (typically 7 years).
• Payment data — not stored; tokenized by processors.
• PII in error logs — scrubbed via Sentry beforeSend hook.

11. Reporting and Cooperation

• NCMEC reporting within 24 hours of detecting apparent CSAM, per 18 USC § 2258A.
• Valid law-enforcement subpoenas processed within 14 days.
• Designated agents: legal@yurivan.com, dmca@yurivan.com.

12. Personnel and Capacity

Current state:

• 1 internal moderator (Trust & Safety lead, founder). NCMEC training in progress; certificate retained on file.
• Automated moderation pipeline: Hive AI Visual Moderation, CSAM Detection, Apparent Age, Celebrity Recognition on every submission and AI-generated image.

Scaling plan:

• At >100 daily submissions: engage TaskUs or Teleperformance Trust & Safety services for 24/7 first-line moderation.
• At >1,000 daily submissions: dedicated full-time T&S analyst.

13. Compliance Calendar

This policy is reviewed quarterly. Triggers for an immediate out-of-cycle review:

• Any NCMEC report filed.
• Chargeback ratio exceeds 0.7% in any week.
• Any law-enforcement subpoena served.
• Material change in Mastercard, Visa, or processor adult content rules.
• New US state or international age-verification statute taking effect.

Contact

• Trust & Safety: trust@yurivan.com
• Mail: Tengen AI Corp, 2224 Augusta Place, Santa Clara, CA 95051 USA