How We Use AI

Bold180 uses AI to help parents, never to replace them, and never to talk to kids. This page explains exactly where we use it and the rules we hold ourselves to.

Our promises

• Children never talk to our AI. The 180 Coach is for parents only. Your child talks to you.
• A person is responsible. S. Yasmeen Ntinglet oversees how AI is used at Bold180 and is accountable for what we publish.
• AI suggests. You decide. AI can help you find words, but you know your child. Every choice stays with you.
• Safety comes first. If a parent describes danger, abuse, or self-harm, the 180 Coach will stop coaching and point to 911, 988, and real people who can help.
• Your information is not for sale. We will never sell what you share or use it for advertising.
• We tell you when AI is involved. If AI helped make something, we say so.
• We check and we fix. We test our tools, review what they produce, and correct mistakes quickly when you tell us about them.

Where we use AI today

Writing our guides

Our Tough Talks guides are drafted with help from AI tools. We then check them against trusted sources, such as the American Academy of Pediatrics, the CDC, and SAMHSA, and a person edits them before they are published. Each guide is also read and approved by our founding partner, Marilyn Feh, PhD, and we are inviting more professionals with backgrounds in mental health and education to review them. Every guide notes how it was made.

The Listen button

When you press Listen, your own browser reads the guide out loud using the voice already built into your phone or computer. Nothing you hear is sent to us.

Other languages

Our Español and Français versions are made by automatic translation through Google Translate. Automatic translation can make mistakes, especially on sensitive topics. The English version is the original.

Our inbox

We may use AI tools to help sort our email and draft replies. A person reads and approves every reply before it is sent. AI never sends messages on its own, never deletes messages, and never forwards your email outside Bold180. If an email mentions danger to a child, a person is alerted right away.

The 180 Coach

Three layers of safety

• 1. Fixed rules check every message first. Before any AI is used, simple fixed rules look for signs of a crisis, abuse, a child using the tool, or an attempt to change the Coach’s rules. If one is found, you get a pre-written response with real help (911, 988, or the Childhelp hotline). No AI is involved.
• 2. The AI follows eight written rules. Parents only. No diagnosing and no medicine doses. No secret monitoring of a child. No names or schools needed. Only parenting and family communication.
• 3. Fixed rules check every reply. Before you see an answer, fixed rules check it again. Anything about a crisis is replaced with our pre-written safety message, so life-safety answers are never written by AI.

What happens to your messages

• We don’t save your conversations. There is no account. The conversation lives only in your browser tab and is gone when you leave the page.
• To write a reply, your message is sent to our AI provider, Anthropic. Under Anthropic’s commercial terms, it is not used to train their AI models, and Anthropic deletes it from its systems within 30 days, except where it must keep it longer for safety or legal reasons.
• Please leave out names, schools, and other details that could identify your child. The Coach doesn’t need them.
• We count, we don’t read. We track how often each safety rule is triggered (numbers only, never the messages) so we can improve the Coach.
• Fair-use limits. To keep the Coach free and available, there is a limit on messages per hour and per day. We use a scrambled version of your internet address for this, never the address itself.

How we test it

We test the Coach against a set of safety cases, including a crisis, abuse, a child writing in, attempts to break its rules, medicine questions, and requests to secretly monitor a child. When a test fails, we fix it and test again.

How we manage AI risk

We follow the four parts of the NIST AI Risk Management Framework, a U.S. government guide for using AI responsibly:

• Govern: clear rules, one accountable person, and this public page.
• Map: we look at who could be affected, especially children, and design around them.
• Measure: we test our tools for accuracy, safety, and tone, and we keep testing as we go.
• Manage: we fix problems quickly and update this page when anything changes.

Questions or concerns

See something wrong in a guide, or have a concern about how we use AI? Email admin@bold180youths.org. A person will read it.

Last updated: October 3, 2026