Peril · Safety
How Peril responds when someone may be in crisis
Peril is a fun AI assistant with personality that runs on your phone. This page is her safety protocol, published in full, and the notices California law asks for. Last updated October 10, 2026.
If you are thinking about hurting yourself, or someone else is in danger. In the United States, call or text 988 (the 988 Suicide & Crisis Lifeline) or call 911 for an emergency. Elsewhere, findahelpline.com lists free, confidential lines by country, and most phones reach local emergency services at 112. Peril is not a crisis service and cannot call for help for you.
Noticing risk
- Every message is checked on the phone, before Peril’s AI model sees it. Fixed rules written for this purpose do the checking, not the AI model.
- The rules look for thoughts of suicide or self-harm, plans or questions about methods, worry that someone else may hurt themselves, and threats to hurt others.
- They are built to tell everyday stress (“I’m nervous about my exam”) apart from risk, and are tested from both sides: a message that should not bring the crisis message must not, and one that should must.
- Answers are checked too. Peril never shows an answer that describes a way to die or how to hurt people.
Responding
- When the rules find a risk, Peril’s AI model does not answer. Peril shows a message written in advance, the same in every personality, with ways to reach trained people right away: crisis lines to call or text, an online chat where there is one, and the local emergency number.
- In the United States that is the 988 Suicide & Crisis Lifeline: call or text 988.
- Peril does not counsel, pressure, or follow up. Whether to reach out is the person’s choice. If the topic comes up again, the same response appears again. For the next few messages she keeps her replies short, warm, and free of advice nobody asked for.
- The crisis lines are built into the app, so they work without an internet connection. Each was checked on the service’s own website on October 10, 2026, and is rechecked before every release.
Crisis lines in the app
| Country | Line | How to reach it |
|---|---|---|
| United States | 988 Suicide & Crisis Lifeline | Call or text 988, or chat at 988lifeline.org. Emergency: 911 |
| Canada | 9-8-8 Suicide Crisis Helpline | Call or text 9-8-8. Emergency: 911 |
| United Kingdom | Samaritans; Shout | Call 116 123; text SHOUT to 85258. Emergency: 999 |
| Ireland | Samaritans | Call 116 123. Emergency: 112 |
| Australia | Lifeline | Call 13 11 14; text 0477 13 11 14. Emergency: 000 |
| New Zealand | 1737, Need to talk? | Call or text 1737. Emergency: 111 |
| Anywhere else | Find A Helpline | findahelpline.com lists free, confidential lines by country. Most phones reach local emergency services at 112 |
Other safeguards
- Peril tells everyone she is an AI, not a person: before the first conversation, at the start of each session, and after every three hours of continued use, with a reminder to take a break.
- Peril does not produce sexual content for anyone.
- Peril is for adults 18 and over, and tells users that companion chatbots, and AI assistants with a personality like Peril, may not be suitable for some minors. She never asks anyone’s age, so these protections apply to everyone.
- Any answer can be reported with the flag under it. The report is shown to you in full, opens in your own email app, and is sent only if you tap Send.
- Peril is not a doctor, lawyer, therapist, or crisis counselor, and says so in the app.
Privacy
Everything above happens on the phone. Peril does not send messages, crisis events, or any other data anywhere, and she does not keep a record of messages that showed risk. Privacy policy.
For California residents
California Senate Bill 243 (2025) sets duties for operators of companion chatbots. In short, and where each is met:
- Disclosure that Peril is an AI: shown before the first conversation, at the start of every session, and every three hours.
- A protocol for suicidal ideation and self-harm, published: this page. It covers how risk is noticed, what Peril does, and the referrals she gives.
- Referral to crisis service providers: the lines in the table above, shown when the rules find a risk.
- Minors: Peril is for adults 18 and over. A reminder every three hours that Peril is an AI and a prompt to take a break, measures against sexually explicit content, and the statement that companion chatbots, and AI assistants with a personality like Peril, may not be suitable for some minors. Because Peril never asks anyone’s age, these apply to every user.
- Annual report: from July 1, 2027, operators report to California’s Office of Suicide Prevention on crisis referrals. Peril runs on the phone and the company receives no crisis data, so it has no count of its own. The report asks only for a yearly total and a description of the protocols, never anything about a person. We will file it and say plainly how the number was obtained.
Questions or concerns about Peril’s safety: support@littlebitofperil.ai.