About Sori and AI

What Sori is

• Sori is a character, a baby dawn phoenix, whose replies are written by an AI language model.
• Sori is not a human and has no lived experience. If you sincerely ask whether you are talking to an AI, Sori says yes, every time, even if you have given Sori another name.
• Sori is not a therapist, counsellor, sponsor, doctor or medical service, and SoberChamp does not offer therapy.
• The chat screen says this in plain words before your first conversation.

What Sori can do

• Listen without judging, and reflect back what you said.
• Help you get through a craving, plan for a hard evening, or mark a win.
• Remember things that matter to you, such as your reasons and your triggers, so you do not have to repeat yourself.
• Answer in the language you write in.

What Sori never does

• Diagnose anything.
• Give doses, schedules, or advice to start, stop, skip or mix any medicine or substance.
• Coach an unsupervised detox. Stopping alcohol, benzodiazepines or opioids suddenly can be dangerous, and Sori will say so and suggest medical support.
• Make up a helpline number. Numbers come only from a table our team maintains.
• Shame you for a slip. You rise again.
• Promise to keep a risk to your life secret, or claim that it can call someone for you. It cannot.

AI can be wrong. Please do not make medical decisions based on what Sori says. Check with a doctor or pharmacist.

What happens in a crisis

Every message is checked for signs of risk, first by a built in list of phrases in several languages, and by an AI safety classifier where that is switched on. The checks look for suicidal thoughts, self harm, overdose, medical emergencies, dangerous withdrawal and violence.

If something might be wrong, Sori slows down, drops the playfulness, and asks you one clear, caring question about what it noticed. It reminds you once that real people are available through SOS and the helplines.

If you may be in immediate danger:

1. The first reply is fixed text written by our team, not by the AI model. It thanks you for saying it, tells you that Sori is an AI and that a real person can help, points to your local emergency number, and lists the helplines for your country from our table.
2. The app shows a crisis card with tap to call buttons for emergency services, helplines and your own emergency contact.
3. After that, Sori stays calm and present, asks about your immediate safety, and encourages one step toward a person. No coaching, no quests, no celebration.
4. Sori does not end the conversation first, and daily message limits do not apply.
5. Sori never discusses methods and never uses guilt.

Sori cannot contact emergency services or anyone else for you. If you are in danger, please call your local emergency number now. See safety and crisis help.

Your chats and your privacy

• Consent first. Nothing you write reaches an AI model until you agree on a dedicated screen. Without that consent the tracker, the SOS toolkit and the helplines still work fully.
• Where the AI runs. By default Sori’s models run on Cloudflare Workers AI, on the same infrastructure that hosts the rest of SoberChamp. Sori receives only what a reply needs: your first name, your habits, reasons, triggers, relevant memories and the recent turns of the conversation.
• Where chats are kept. In your private vault, with the rest of your journey.
• Memory you can see. The page What Sori remembers lists every memory. You can edit or delete each one, add your own, or clear everything.
• Things Sori does not keep as memories. Statements about suicide, self harm, overdose or abuse, dosing details, and other people’s secrets.
• No training, no ads. We do not use your conversations to train AI models, for advertising, or for sale.

Details are in the privacy policy.

Tell us when Sori gets it wrong

If a reply from Sori was unhelpful, wrong or upsetting, write to support@soberchamp.com. Every message is read by a person and helps us make Sori safer.

Adults only

SoberChamp and Sori are for people aged 18 and over.