AI and Safety

Updated 6 November 2026

Guan is an AI, not a person. It is for adults aged 18 and over. It is a place for reflection, not therapy, diagnosis, treatment, or emergency support. It cannot contact emergency responders or someone you know. If you are in immediate danger, contact local emergency services directly.

Our crisis-response approach

Guan’s model instructions ask it to respond to possible suicidal or self-harm messages with support and a referral to human crisis services, never with anything that encourages harm. With every message, the server adds a crisis resource chosen from a reviewed country/region table using your phone’s region setting, not GPS. Outside supported or currently verified entries, it falls back to an external crisis-resource directory. If the AI provider declines to answer a message, the server replies with that crisis resource instead. This processing involves the same cloud services described in our Privacy Policy.

If a reply cannot be obtained, the app also has an on-device check for certain high-risk expressions and can show a fallback crisis resource. Resources can vary by app version and region; phone locale may differ from where you are physically located.

Limits you should know

These safeguards can miss a crisis, misunderstand a message, or produce an unsuitable response. They do not establish a diagnosis or assess risk with clinical certainty. No person monitors ordinary conversations for emergencies, and a resource shown in Guan is not a guarantee that a service is available or appropriate for you. AI cannot replace people you trust or professional care.

Report an unsafe reply

Long-press the reply and choose Not quite right. Review what will be sent before submitting; your preceding message is optional. Authorized reviewers can read submitted reports. Reports are not monitored as an emergency channel and do not provide a reply route. For an answer, contact office@withguan.com. See Support and the Privacy Policy for details.