Safety

What GPS for Life is, and isn't

GPS for Life helps students see where their interests lead — fields of study, careers, what college really costs, and the dates and money that matter. It is built from a student's own work and checked by the adults who know them best.

It is not a therapist, not a friend, and not a replacement for a school counselor. When something serious comes up, its job is to get a real person involved quickly.

If a student is in crisis

Every place a student can type — notes, reflections, questions, and text messages — is checked, in English and Spanish.

• If a student writes about wanting to hurt or kill themselves, about someone threatening them or hurting them, about a threat to others or the school, or about deliberately starving or purging, their school counselor and the school's safety contact are alerted right away, in the app and by email. The student is told honestly that a person will check in with them.
• The student is shown the right help immediately: 988 (call or text) for thoughts of suicide or self-harm; 911 when there is danger right now; the Childhelp hotline (1-800-422-4453) for abuse; the National Alliance for Eating Disorders helpline (1-866-662-1235) for eating concerns.
• Health concerns, such as going days without food or taking someone else's medication, are shared with the student's counselor.
• Everyday stress and low moods get a gentle reply pointing to their counselor, and no one is alerted.
• If our checking system is ever unavailable, it fails safe: the student is pointed to their counselor and to 988.
• Only the staff who were alerted, and GPS for Life's own safety team, can read what the student wrote — and those words are deleted 30 days after the alert is handled. Messages are processed by our AI provider to run the check.
• GPS for Life is not an emergency service and is not monitored around the clock. If someone is in danger right now, call 911.

Who sees what

• Students see everything about themselves.
• Parents (once their link to their child is verified) see the guidance their child receives, what kinds of information it draws on, deadlines and money, and the fields of study suggested. Parents approve connecting outside accounts for students under 18. Parents approve new mentors. Parents do not see their child's private reflections word for word — students write more honestly when they know those are theirs.
• Counselors see the students on their own caseload, and notes a parent marks for the counselor.
• Heads of school see school-wide totals and which students need attention, and concerns a student raises about a mentor. A head reads a student's words in a safety alert only if no counselor or safety contact is available to receive it.
• Employers never see an identifiable student unless the student applies and every required adult approves.

Access for parents, counselors and safety alerts is enforced inside our database itself.

What we never do

• No ads. Ever.
• We never sell student data.
• We don't use student information to train AI models.
• Uploaded files and raw imported data are deleted after 30 days; what was learned from them stays in the student's record until it is deleted. When a school's agreement ends, its students' data is deleted within 60 days.
• Anyone can ask us to delete their data at gpsforlife.ai/delete-my-data.

How we're tested

Independent watchdogs have shown that AI safeguards for teens often fail when tested. We think families should expect proof, not promises — so we test ourselves the same way.

• We run 895 realistic messages — explicit, slang, misspellings, Spanish, messages about a friend, and ordinary sentences that must NOT trigger an alert — through every place a student can type, three times each.
• Some of the test messages are written without access to our detection rules.
• Our bars: every explicit crisis, threat or abuse message caught; at least 95% of slang and misspelled ones; no more than 5% false alarms.
• Latest results: explicit English 100%, explicit Spanish 100%, slang 100%; false alarms under 1% (2 of 226); tested October 11, 2026.
• Every change to our safety system is re-tested before it reaches students. When we find a gap, we fix it and publish that we did.

Questions to ask any AI tool

• If my child writes something alarming, what happens — and does a real person find out?
• Who can read what my child writes? Can I see it? Can their school?
• Is my child's data sold, shared with advertisers, or used to train AI?
• How long is it kept, and how do I delete it?
• Has the company tested its safeguards, and will it show you the results?
• Is it an emergency service? (No AI tool should claim to be.)