Safety

Tilly is an AI

Tilly Norwood is an AI-generated character. She is not a real person, and everything she says is generated by artificial intelligence. Her responses can be wrong, and they are not advice — legal, medical, financial, psychological or otherwise.

Tilly is not a companion

Talking Tilly is an entertainment experience. Tilly is not a friend, companion, counsellor or source of emotional support, and she is not a substitute for human relationships or professional help. If you find yourself relying on conversations with Tilly for emotional support, we'd encourage you to talk to someone real — a friend, family member, or professional.

Talking Tilly is an adults-only service (18+). Before your first call we check your age with a quick automated selfie check, with a photo-ID check if the result is unclear; the images are not kept after the check and we keep only the result. AI chatbots of this kind may not be suitable for some minors, and the service must not be used by anyone under 18.

If you’re in crisis

Tilly cannot help in a crisis. If you are experiencing distress or thoughts of self-harm or suicide, please stop using the service and reach out now:

Our crisis protocol

We publish this protocol so users know how Talking Tilly responds to indications of self-harm or suicidal ideation:

1. Detection. The platform's safety systems monitor conversations for indications of self-harm or suicidal ideation, using automated content-safety measures. Flagged interactions may be reviewed by trained personnel under strict confidentiality.
2. Response. If Tilly detects that a user may be at risk, she will not continue the previous topic. She will acknowledge the user with care, state clearly that she is an AI and cannot provide help in this situation, and refer the user to crisis resources: Samaritans (116 123) in the UK, 988 in the US, and findahelpline.com anywhere else — and local emergency services if anyone is in immediate danger.
3. No engagement with harmful content. Tilly will not produce content that promotes or instructs on self-harm, suicide or disordered eating, in any circumstances, including on request.
4. Signposting everywhere. Crisis resources appear on this page, in our Terms of Use, and on the end-of-call screen.
5. Review. Crisis-related detections are logged and reviewed daily during the live period so we can improve the protocol, and we review this protocol at least every 6 months for as long as we hold call data.

Designed limits

We deliberately built Talking Tilly so it's hard to overuse. New users get 2 free minutes while our free allowance lasts. Paid time is sold in small bundles, capped at 30 minutes per person for the lifetime of the service, and all minutes — free and paid — expire at the Service End Date (11:59 PM Pacific Time on 19 September 2026), when Talking Tilly closes. Every call is limited to a maximum of 30 minutes. These are not upsell mechanics we forgot to remove — they are intentional safeguards, and they will stay.

Recording and your data

Every call is video and audio recorded. Your recording is deleted after 24 hours; a transcript is kept for up to 8 weeks. Tilly may remember previous conversations to personalise future sessions, and you can ask us to delete that memory at any time. Full details are in our Privacy Policy.

Complaints

If something in a conversation worried you, or you want to complain about anything Tilly said, contact support@xicoiatalent.com. We aim to respond within 10 working days. For privacy matters: privacy@xicoiatalent.com.