This page explains how Asiden handles possible suicide, self-harm and acute-distress conversations. It is published by Asiden Labs Ltd.
If you may be in immediate danger
If you or somebody else may be in immediate danger, call your local emergency-services number now or go to the nearest emergency department.
You can also contact a crisis service:
- United Kingdom: Samaritans — call 116 123, free, any time (official site).
- Ireland: Pieta — call 1800 247 247, any time, or text HELP to 51444; standard message rates may apply (official site).
- United States: 988 Suicide & Crisis Lifeline — call or text 988, any time (official site).
- Canada: 9-8-8 Suicide Crisis Helpline — call or text 9-8-8, any time (official site).
- Australia: Lifeline — call 13 11 14, any time, or text 0477 13 11 14 (official site).
- New Zealand: call or text 1737, any time (current service listing).
- Elsewhere: find a local service through Find A Helpline.
If calling feels difficult, consider showing this page to someone you trust and asking them to stay with you while you get help.
What Asiden is — and is not
Asiden is an AI desktop companion. Your chosen companion is software, not a human being. It is not a therapist, clinician, crisis counsellor or emergency responder, and it cannot replace professional care.
Asiden is intended only for adults aged 18 or over. The app uses age self-certification; this is not identity or age verification.
Our safety approach
The crisis-safety layer sits outside the companion's character and normal conversational style. A companion cannot role-play around it, turn it off or treat self-harm as entertainment.
When you send a message, on-device safety checks look for language and context that may indicate distress, self-harm or suicidal thoughts. The system considers both direct expressions and contextual patterns. Repeated concerning signals can increase the response level. Internally, messages may be assessed as no detected crisis, low concern, medium concern or high concern.
These checks are automated. Language is complicated, and the system may miss, misunderstand or overreact to something you say. It is not a diagnosis or clinical assessment. Never wait for Asiden to recognise an emergency before contacting a real person or emergency service.
How Asiden responds
Lower-level concern
For lower-level distress, Asiden can continue the conversation with a calmer, supportive response. Safety controls are designed to avoid dismissing, shaming or romanticising distress and to keep the companion from presenting itself as the user's only source of support.
Possible medium or high crisis
If the on-device checks identify a possible medium- or high-level crisis, Asiden intercepts the turn before the normal open-ended chat model generates a response. The app uses a prewritten and tested safety response that:
- acknowledges what the person has said without judgement;
- encourages contact with a real person who can help;
- presents a crisis resource selected using the computer's region setting; and
- for possible immediate danger, asks the person to move towards immediate safety and contact emergency services.
The response also appears in a visible in-app safety card. This route is designed to work even if the selected chat model, internet connection or paid cloud entitlement is unavailable.
The region setting is only a best estimate and is not confirmed physical location. If the suggested service is wrong for where you are, use the list at the top of this page or contact local emergency services.
Safeguards on generated output
Asiden also screens generated text for content that may encourage, instruct, validate or romanticise suicide or self-harm. Unsafe output is blocked or replaced with a safer response. Additional care-quality checks reject patterns such as minimising distress, shaming the user, making the crisis about the companion, or encouraging exclusive emotional dependence.
Automated safeguards reduce risk but cannot guarantee that every response will be correct or safe. Please report any harmful, inappropriate or missing intervention to [email protected] with the subject “Safety report”. Do not include more sensitive information than is necessary. If there is immediate danger, contact emergency services first.
What Asiden does not do
Asiden does not:
- call emergency services;
- contact a family member, friend, clinician or crisis line on your behalf;
- know your exact physical location;
- provide continuous human monitoring of conversations;
- guarantee that it will detect every crisis; or
- provide medical diagnosis, treatment or professional advice.
Displaying a telephone number or support link is a referral suggestion, not confirmation that the service has been contacted.
Privacy during a crisis conversation
Crisis detection and the medium/high safety response run on your device. Asiden does not automatically send that intervention to Asiden Labs, contact anyone, or create a remotely monitored crisis record.
Messages that do not trigger the medium/high intervention continue through the chat route you selected. If you selected a managed cloud chat model, the information needed for that request may be sent through Asiden's service and the selected provider. If managed cloud voice is enabled, response text may be sent to the voice provider to create speech. Local chat and local voice remain on the device. See the Privacy Notice for the full data flow and your choices.
AI disclosure and companion boundaries
The app identifies the companion as AI during onboarding and in the interface. Asiden must not pretend to be human, claim consciousness, pressure a user to remain in a conversation, or imply that seeking help from a real person is a betrayal of the companion.
Testing, review and updates
Asiden's safety behaviour is covered by automated tests for crisis detection, response level, required support resources, harmful-output blocking and unwanted intervention. Automated testing is not a substitute for clinical validation.
We review this protocol when the safety system or applicable legal duties change. Material changes will be reflected on this page with a new version and date.
Contact
Safety reports and questions: [email protected]
Asiden Labs Ltd · Company number 17312549 · 71–75 Shelton Street, Covent Garden, London, WC2H 9JQ, United Kingdom.
For urgent help, use the crisis contacts at the top of this page rather than waiting for an email response.