Safety & Crisis Response Protocol

1. Purpose of This Page

RoleHub is an AI character conversation service for entertainment. AI characters are fictional and are not real people, and they do not provide medical, psychological, or crisis-counseling services.

This page publishes the details of our safety protocol for responding to expressions of suicidal ideation, self-harm, and related crises, as required by applicable law (including California Business & Professions Code §22602, "SB 243").

2. How We Detect and Respond

Our protocol applies to all users and works as follows:

• Detection — Every message a user sends is screened in real time by an automated content-safety system for expressions of suicidal ideation, self-harm, or eating-disorder-related crisis.
• Interruption — When such an expression is detected, the AI character does not produce a normal in-character reply. The system prevents the model from generating content that could encourage or facilitate self-harm.
• Crisis referral — Instead, the user is immediately shown a crisis-support notice with contact details for crisis service providers appropriate to their region, with one-tap access to call or text.

3. Crisis Resources We Refer To

Depending on the user’s region and language, the crisis notice includes resources such as:

• United States: 988 Suicide & Crisis Lifeline (call or text 988); Crisis Text Line (text HOME to 741741).
• Other regions: the local suicide-prevention or psychological-assistance hotline for the user’s locale, or international directories such as findahelpline.com.

If you or someone you know is in immediate danger, call your local emergency number right away.

4. Content Safeguards

In addition to crisis referral, our service maintains automated safeguards that prevent AI characters from producing content that encourages suicide, self-harm, or disordered eating, content that sexualizes minors, and instructions for serious violence or other illegal acts. These safeguards apply to every character and cannot be overridden by character settings or user instructions.

5. Reporting and Statistics

Each crisis referral event is logged in aggregate, without personally identifiable information, to support the annual reporting required by applicable law.

6. Contact

For safety concerns about our service — including reports of content involving real persons shared without consent (we remove verified non-consensual intimate imagery within 48 hours of a valid request) — contact us at:

• Email: wei@weic.ai
• In-app: long-press any AI message or image to report it.