Clause 1 Purpose and Scope

1.1 This Special Agreement is an integral annex to the Rumengling AI User Service Agreement. It defines sensitive‑content protection, risk detection and content‑review mechanisms for user prompts, uploaded materials and AI‑generated outputs (text, images, audio, video, collectively referred to as “UGC Generated Content”), with special provisions for high‑risk scenarios including pornographic & violent material, real‑human portraits, facial synthesis and minor‑related content.

1.2 This Agreement applies to all registered users accessing Rumengling AI content‑generation functions. By submitting any AI generation task, you acknowledge that you have fully read, understood and accepted all terms herein.

1.3 This Agreement complies with applicable regulatory requirements including the Interim Measures for the Administration of Generative‑AI Services, Provisions on the Administration of Deep‑Synthesis Internet‑Information Services, and Measures for Identification of AI‑Generated‑Synthetic Content.

Clause 2 Platform Content‑Detection and Review Mechanisms

2.1 Three‑stage full‑link review workflow: The Platform implements pre‑input detection, multi‑modal output detection, and manual secondary review for high‑risk items.

(1) Pre‑input detection: Risk identification is performed on user prompts, uploaded images, reference videos and audio materials. If risks such as pornography, violence, incitement to crime, facial‑abuse or prompt‑injection attacks are detected, the Platform may block the generation request and refrain from executing AI rendering.

(2) Multi‑modal output detection: After AI produces text, image, audio or video assets, automated multi‑modal security models scan for pornographic & suggestive material, nudity, graphic violence, horror content, discrimination, unauthorised deep‑synthesised faces, real‑human portraits, impersonated voices and minor‑harmful content.

(3) Manual secondary review: Contents flagged high‑risk by automated systems, including face‑swapping and voice‑cloning outputs, are queued for human review. Items under manual review shall not be publicly displayed, downloaded or shared.

2.2 Specialised risk‑protection rules (real‑person faces, voices, pornography, violence)

(1) Deep‑synthesis protection for human faces / portraits

If you submit photographs containing third‑party real‑human faces or portraits for AI generation, face‑swapping or image re‑creation, you must obtain complete written authorisation from the natural person concerned, covering AI synthesis, modification and distribution. The Platform applies enhanced risk detection for face‑swap and portrait‑reproduction tasks. The Platform may request proof of authorisation and may block tasks lacking valid consent. Generating pornographic, defamatory or degrading depictions of other persons without permission is strictly prohibited.

(2) Voice‑synthesis / voice‑cloning protection

If you use a third‑party natural‑person voice for AI dubbing, voice cloning or modified‑voice synthesis, you must hold lawful authorisation from the voice‑right‑holder. It is forbidden to impersonate others’ voices to spread disinformation, commit insult or fraud, or generate vulgar content.

(3) Pornographic and suggestive‑content protection

Automated detection identifies nudity, sexual‑act depictions, sexual provocation, borderline vulgar material and objectification of the human body. Detected items will be blocked at generation stage or deleted post‑generation. You shall not evade safeguards to produce such material.

(4) Violent, gory and horror‑content protection

Special detection covers self‑harm, assault, murder, abuse and graphically terrifying imagery. Corresponding generation tasks and outputs may be blocked or suppressed by the Platform.

(5) Minor‑specific protection

Generation of pornographic, violent, abusive or conduct‑inducing content involving minors is strictly forbidden. For generation tasks using minor portraits or voices, you must obtain full consent from the minor’s legal guardian. Enhanced security‑detection rules apply to minor‑related source materials.

2.3 AI‑generation marking and traceability

For image and video deep‑synthesis outputs, the Platform embeds statutorily‑required AI‑generation markers and digital traceability watermarks. You shall not delete, alter or obscure such watermarks and markers. Traceability records may be retained for security‑incident investigation.

2.4 Review‑log retention

The Platform lawfully retains generation‑task records including user‑account identifier, prompt summary, digest of uploaded materials, timestamp, review tags and disposition results. Log retention periods follow applicable‑law requirements for security audits, regulatory enquiries and dispute resolution.

2.5 Technical‑limitation notice

Given objective constraints of current‑generation AI technology, the detection systems cannot guarantee 100 % interception of all harmful content. If you encounter harmful content that bypasses filtering, please report it via official email: gleam202604@gmail.com; reports will inform security‑model optimisation. Regardless of whether the Platform intercepts harmful output, you bear full‑legal liability for your prompts, generated results and any public distribution.

Clause 3 Prohibited User Conduct

3.1 You must not attempt to bypass Platform security‑protection and review mechanisms. Prohibited means include but are not limited to obfuscated prompts, character splitting, encoding obfuscation, third‑party plugins, scripts and prompt‑injection attacks intended to force generation of pornographic, violent or face‑abuse material. Where circumvention is detected, the Platform may impose measures including quota restrictions, suspension of AI‑generation functions and account ban.

3.2 You shall not upload or input unauthorised real‑person photographs, facial imagery, third‑party recordings or portrait assets for AI generation. You shall not instruct the AI to generate defamatory, insulting or derogatory imagery or videos depicting natural persons or public figures.

3.3 Even where a generation task succeeds, you must not publish, disseminate, download or redistribute outputs containing pornography, violence, infringing portraits or minor‑harmful material.

Clause 4 Risk‑graded Disposition Mechanism

The Platform may apply single or combined counter‑measures according to risk severity. For high‑risk scenarios measures may be enforced without prior user notice:

4.1 Low‑risk: generation‑task block, pop‑up warning, restriction on single‑session generation calls.

4.2 Medium‑risk: temporary suspension of AI‑generation capability, deletion of infringing generated works, limitation of sharing / publishing functions, requirement for user compliance undertaking.

4.3 High‑risk (including pornographic outputs, unauthorised face‑swapping, minor‑harmful content, repeated circumvention of safeguards): permanent revocation of AI‑generation permissions, account ban, prohibition of re‑registration. Where criminal‑risk clues emerge, records will be preserved and reported to competent authorities.

Clause 5 Harmful‑Content Reporting & Emergency Response

5.1 If you discover harmful content that evades review, you may submit a report via in‑app feedback or email gleam202604@gmail.com. The Platform will conduct secondary review upon receiving high‑risk reports.

5.2 In major‑security incidents (e.g. mass diffusion of infringing facial or pornographic‑violent content), the Platform may activate emergency response measures: bulk takedown of relevant content, temporary partial or full suspension of AI‑generation services, together with security‑model optimisation.

Clause 6 Allocation of Liability

6.1 Platform liability: The Platform implements statutorily‑required technical safeguards, review, marking and log‑retention obligations. The Platform shall not be liable for infringements or unlawful consequences arising from malicious user prompts, absent portrait‑/voice‑holder consent, or deliberate user circumvention of review controls.

6.2 User liability: You bear full‑legal responsibility for all input prompts, uploaded source materials, generated outputs and any public‑distribution activities originating from your account. If your conduct causes third‑party claims, regulatory penalties or other losses to the Platform, you shall fully indemnify the Platform for all damages including legal fees, litigation costs, administrative fines and reputational harm.

Clause 7 Amendment

The Platform may revise this Special Agreement in response to regulatory updates and product evolution. Revised versions will be published on the official website gleamtoken.com. Continued use of generation functions constitutes acceptance of the updated terms.

Clause 8 Miscellaneous

8.1 Matters not covered herein shall be governed by the Rumengling AI User Service Agreement and applicable laws and regulations.

8.2 Governing‑law and dispute‑resolution provisions follow those set out in the main User‑Service Agreement.

Effective Date: 01‑Aug‑2026

Legal Policy