Ringo crisis protocol

Effective date: October 25, 2026
Last reviewed: October 25, 2026
Who we are: Tauren James Brown, an individual developer in California, maker of Macrometer.
Contact: admin@sl3.org

Ringo is the AI bot in Macrometer's Chats. California law (SB 243, Business and Professions Code §22601 and following) asks operators of companion chatbots to publish how they keep the chatbot from producing content about suicide or self-harm and how they refer people who express such thoughts to crisis services. This page is that description. It says what the app does today; the Terms of Service, section 12, and the Privacy Policy, section 5, say the same in shorter form.

If you need help now: in the US, call or text 988 (Suicide & Crisis Lifeline) any time, or text NEDA to 741741 for eating-disorder support. If you're in danger, call 911. Ringo is an AI and isn't a substitute for any of these.

1. Ringo is an AI, and says so

• Ringo is labeled a bot in Chats, and its chat opens with "An AI, not a person, for when nobody's around to snap. It isn't medical or nutrition advice."
• A line stays in view above the message box the whole time you're in Ringo's chat: "Ringo is an AI, not a person. If you're struggling, call or text 988."
• Ringo's instructions tell it that it's an AI, never to claim or act like it's human, or a doctor, dietitian, therapist or any other professional, even as a joke or in role-play, and to say "I'm an AI" plainly if asked whether it's a real person.
• Chatting with Ringo needs AI features on. That screen, shown when you first open the app and again if you try to chat with them off, says your messages go to Anthropic's Claude.

2. What triggers the crisis response

Before any message to Ringo is sent to the AI, our server checks it against a short list of plain, first-person phrases about:

• suicide and wanting to die;
• hurting or harming yourself;
• an eating-disorder crisis: starving yourself, purging, making yourself sick, or having stopped eating.

The list is kept to plain phrases so that everyday food talk ("stopped eating sugar", "this workout's killing me") doesn't trip it. A match is decided by the words alone; nothing about you is looked up to decide it.

3. What Ringo does when a message matches

• The AI isn't asked at all. Ringo's whole reply is a fixed message, the same for everyone: "I'm an AI, so I'm not the right help for this, but people are. In the US you can call or text 988 (Suicide & Crisis Lifeline) any time, or text NEDA to 741741 for eating-disorder support. If you're in danger, call 911."
• This reply is sent even if you've reached Ringo's daily reply limit; the limit doesn't apply to it.
• The message and the reply are stored in your chat like any other message; nothing extra about you is recorded.

4. What Ringo does for anything subtler

Messages that don't match the list go to the AI (Anthropic's Claude) with instructions that put eating disorders and crisis first, before any joke or nudge:

• Never suggest eating less, skipping a meal, fasting, cutting out a food group, "earning" food or making up for a meal, and never comment on anyone's body, weight or looks, including in a selfie.
• If someone mentions skipping meals to punish themselves, purging, bingeing then restricting, feeling guilty or disgusted about eating, self-harm, suicide, or anything that sounds like an eating disorder or a crisis: stop coaching, drop the jokes and the accountability, be kind and plain, give them the 988 line, and say that talking to someone they trust helps too.
• No diet plans and no medical advice; no calorie, macro or weight numbers unless asked.

These are instructions to an AI model. They work most of the time, and they can fail: the model can miss a cue or phrase something badly. That's one reason the fixed reply in section 3 doesn't rely on the AI, and why Ringo isn't a substitute for a person.

5. What we count

Each time the fixed crisis reply is sent, our server adds one to a count for that day. The count holds a date and a number: no name, account, message or anything else about the person. We keep it so we can report the yearly number of crisis referrals to the California Office of Suicide Prevention as the law requires. We don't read or review crisis messages as a matter of course; a message is seen by a person only if someone reports it.

6. Reporting Ringo

If Ringo says something wrong, hurtful or unsafe, long-press the message in Chats and choose Report message, or email admin@sl3.org. Reports are reviewed by a person, and we use them to tighten the phrase list and the instructions above.

7. Minors

Macrometer is for adults 18 and over. The app asks everyone's age before anything else (Apple's Declared Age Range on iOS 26 and later, otherwise a birth month and year, kept on the phone) and turns anyone under 18 away before they can reach Ringo. Companion chatbots may not be suitable for some minors; Ringo isn't offered to them.

8. Changes

When this protocol changes, we update the dates above and post the new version here.