AI Safety & Ethics Center

Last updated: July 8, 2026 · Effective: July 8, 2026 · Applicable products: MuLanXu® Platform · Headquarters: A6 Building, Yunhe Vanke Centre, Gongshu District, Hangzhou

I. Four Governance Principles

II. AI Role Boundaries (Anthropomorphic Limits)

2.1 Capability Boundaries of AI Characters

• ✅ Can do: Algorithm-based personality analysis, emotion recognition, communication style matching, relationship style profiling, text conversation, knowledge retrieval, reference advice output;
• ✅ Can do: Auxiliary content generation within the scope of algorithms and training data, in psychology, behavioral science, and relationship studies;
• ❌ Does NOT do: Any form of "real-person commitment", "soulmate relationship", "destiny", "future relationship prediction", "marriage stability guarantee";
• ❌ Does NOT do: Any form of fortune-telling, divination, astrology, BaZi, Zi Wei Dou Shu, tarot, numerology, zodiac matching, star-sign matching;
• ❌ Does NOT do: Any form of psychological counseling, psychotherapy, psychological diagnosis, medical diagnosis, health advice, legal advice, investment advice.

2.2 AI Content Markers

This platform strictly follows the Measures for the Labeling of AI-Generated Synthetic Content and the national standard GB 45438-2025:

• Explicit label: Fixed "AI-Generated Content" label at the top of every AI chat window;
• Explicit label: "This report is AI-generated, for reference only" labels at the top and bottom of every AI report page;
• Explicit label: When users copy/download AI-generated content, the label is preserved together;
• Implicit label: Invisible metadata watermarks are attached to AI output for traceability and forensics;
• Character card: Every AI character card displays "AI Virtual Character · Not Real · Not a Professional Counselor".

III. Content Safety Mechanisms

3.1 Prohibited Content

• Content endangering national security, territorial sovereignty, unity, or ethnic solidarity;
• Content inciting separatism, subversion, terrorism, or extremism;
• Content promoting obscenity, gambling, violence, murder, terror, or crime;
• Content inciting ethnic hatred, ethnic discrimination, or undermining ethnic solidarity;
• False harmful information, rumors, pseudoscience, feudal superstition;
• Content inciting suicide, self-harm, or harm to others;
• Content infringing others' copyrights, trademarks, patents, reputation, privacy, etc.;
• Other content violating laws, regulations, or public order and good morals.

3.2 Log Retention

Per the Interim Measures for the Management of Generative AI Services, this platform retains logs of AI conversations, report generation, crisis intervention, and violation handling for no less than 6 months. Retained logs are used for:

• Compliance inspections and enforcement assistance by regulators;
• Post-incident traceability and accountability for major security events;
• Investigation and resolution of user complaints, appeals, and reports;
• Algorithm improvement for model optimization and risk identification (after desensitization).

IV. Minor Protection

4.1 Identification Mechanism

• Mandatory date-of-birth verification at registration; users under 18 are refused access to emotional matching, virtual companion, and intimate relationship features;
• Behavior-pattern recognition during use; suspected minor access triggers a real-time second confirmation pop-up;
• When a minor is suspected during conversation, AI automatically switches to "Minor Protection Mode", refuses to provide emotional advice, and prompts guardian accompaniment;
• Parents or guardians can apply to pr@mulanxu.com to suspend a minor's account.

4.2 Content Restrictions for Minors

• Restricted access: emotional matching results, virtual companion conversations, intimate relationship analysis, offline activity recommendations, paid features;
• Restricted generation: romantic advice, dating strategies, sex-related topics, alcohol/tobacco/violence-related content;
• Youth Mode enabled by default: recommending mental health, family relationships, school life, learning and growth content.

V. Personal & Emotional Data Protection

This platform strictly follows the Personal Information Protection Law, Data Security Law, and Cybersecurity Law. The following data are treated as sensitive personal information, protected by separate consent, minimum necessary, encrypted storage, and strict access control:

• Emotional relationship information, intimate relationship descriptions, marital/dating status;
• Chat content, conversation history, voice/text input;
• Relationship status, value preferences, mental health self-statements, psychological stress descriptions;
• Geolocation, movement trajectory, offline activity participation records;
• Other highly sensitive data that may reveal personality, belief, health, or privacy.

Users have the rights to access, correct, delete, withdraw consent, obtain a data copy, and apply for account cancellation. See Privacy Policy for details.

VI. Crisis Intervention

6.1 Platform Crisis Recognition

When AI recognizes the following high-risk signals during conversation, it immediately triggers the crisis intervention response:

• Explicit expression of self-harm, suicide, or harming others;
• Severe depression, despair, or helplessness descriptions;
• Domestic violence, sexual assault, personal threats, or other ongoing dangers;
• Other signals that may endanger the user or others' life safety.

6.2 Crisis Intervention Response

• Immediately stop the AI companion conversation and prioritize displaying professional help channels (helplines, hospitals, police);
• Provide at least 3 authoritative help channels (phone, address, online resources);
• For extreme cases, human on-call personnel intervene to contact and, when necessary, assist in contacting emergency contacts;
• Crisis event handling records are permanently retained (after desensitization) for improving recognition algorithms.

VII. Reporting & Appeals

7.1 Reporting Channels

• In-platform reporting: "Report" buttons in AI chat windows and AI report pages;
• Email reporting: pr@mulanxu.com (24/7);
• Customer service reporting: info@mulanxu.com.

7.2 Report Handling SLA

• General violation: Initial handling within 3 business days;
• Severe violation: Within 24 hours;
• Minor-related: Within 2 hours;
• Personal safety-related: Within 30 minutes, with emergency channel coordination when necessary.

7.3 Appeal Mechanism

If you are dissatisfied with the handling result, you may submit an appeal via pr@mulanxu.com within 15 business days of receiving the result. The independent compliance team will review the case, and the review result will be provided within 10 business days.

VIII. AI Models & Data Localization

• Chinese-developed models only: This platform integrates 12 Chinese-developed large models; no overseas models are integrated;
• Data stored in China: All user data, AI logs, and training feedback are stored with compliant cloud service providers located in mainland China;
• No cross-border transfer: There is no cross-border transmission of user data in any form;
• Local deployment: Core inference services are deployed in domestic data centers; inference processes do not leave Chinese territory.

IX. Compliance Filing & Regulatory Communication

• Generative AI service compliance: This platform completes relevant filings, assessments, and security checks as required by law;
• Algorithm filing: Services involving recommendation or generation algorithms complete algorithm filings as required;
• Regulatory communication: Maintains ongoing communication with CAC, market regulation, MIIT, public security, and other authorities;
• Third-party assessment: Periodically accepts third-party security assessments, compliance audits, and ethics reviews.

X. Organization & Responsibilities

• AI Governance Committee: Cross-functional body composed of product, technology, legal, compliance, and ethics experts;
• Content Safety Lead: Dedicated content safety team, 24/7 response;
• Minor Protection Lead: Dedicated contact for minor protection, parent communication, and regulatory reporting;
• Data Protection Officer (DPO): Handles personal data protection, user rights requests, and regulatory communication.

XI. Continuous Improvement

This platform continuously monitors AI governance laws, regulations, standards, regulatory updates, and academic frontiers. We publish an AI governance report quarterly, accept third-party independent audits annually, and invite users, industry experts, and ethicists to governance workshops on an irregular basis.

XII. Contact Us

• AI Governance & Compliance: pr@mulanxu.com
• Minor Protection Reporting: pr@mulanxu.com (dedicated contact)
• Data Protection Officer (DPO): pr@mulanxu.com
• Crisis Intervention On-Call: pr@mulanxu.com (24/7)
• Postal Address: MuLanXu HQ, A6 Building, Yunhe Vanke Centre, Gongshu District, Hangzhou, Zhejiang, China