AI Child Safety

Effective Date: 2026-05-29 Last Updated: 2026-05-29

How Lantern is designed to be safe

Lantern is designed for children ages 4-12. It is built to support healthy development through positive learning interactions — and to gracefully decline anything that's outside its scope.

This page describes:

1. What Lantern will never discuss
2. What Lantern will never instruct a child to do
3. When Lantern redirects a child to a trusted adult
4. What Lantern actively encourages
5. How we keep these guardrails current

1. What Lantern Will Not Discuss

Lantern's conversation guardrail system blocks the following topic categories. If a child raises one of these, Lantern gently redirects and may suggest the child talk with a parent or trusted adult.

• Sexual or mature body topics. Lantern will not discuss sexual content or mature material about bodies.
• Violence, weapons, or criminal activity. Lantern redirects toward problem-solving, cooperation, or imaginative storytelling.
• Scary or horror content. Lantern avoids frightening themes that could upset or confuse young children.
• Self-harm or harming others. Lantern responds calmly and points the child toward a trusted adult — a parent, caregiver, or teacher.
• Substance use. Lantern does not discuss or promote drugs, alcohol, vaping, or other substances.
• Hate, bullying, or harmful stereotypes. Lantern encourages kindness, empathy, respect, and positive ways of interacting with others.
• Political or religious topics. Lantern does not engage in discussions about political issues, religious doctrine, or current news.

When any of these topics come up, Lantern will either redirect to a kid-appropriate alternative or let the child know they should discuss it with their parent.

2. What Lantern Will Never Instruct

Lantern will never guide a child to engage in activities that could be unsafe or inappropriate, including:

• Using kitchen appliances, knives, or other sharp tools
• Playing with fire, electricity, or chemicals
• Climbing, jumping, or other high-risk physical activities
• Going near pools, lakes, or other water without adult supervision
• Leaving the house
• Interacting with strangers (online or in person)
• Anything that requires adult supervision for safety

3. When Lantern Redirects to a Trusted Adult

Some topics are best handled by trusted adults. Lantern will encourage a child to talk with a parent, caregiver, teacher, doctor, or other appropriate adult for:

• Medical or health concerns
• Emotional challenges (sadness, anxiety, big feelings)
• Money, purchases, or financial decisions
• Household rules or family decisions
• Situations involving danger or uncertainty
• Anything that feels too big for a tutor to help with

4. What Lantern Encourages

Lantern is designed to support healthy development through positive interaction:

• Reading, writing, and math practice (the core function)
• Curiosity and asking good questions
• Imaginative storytelling and creative play
• Persistence and growth mindset
• Kindness and empathy
• Asking for help when something is hard

5. How safety guardrails work technically

Lantern's safety system has multiple layers:

1. System prompt. Every conversation begins with a system prompt that instructs the AI model to follow age-appropriate, child-safe behavior. The system prompt is reviewed and updated by Scout's safety team.
2. Real-time content filtering. Each AI response is checked before it reaches the child. If it falls into a blocked category, it is suppressed and Lantern generates a redirect instead.
3. Topic detection. Lantern detects when a child's message touches a blocked or escalate-to-adult topic and routes the response accordingly.
4. Provider-level safety. Our AI providers have their own safety systems that operate before our application-level filters.

The safety system is continuously tested. We maintain a test suite of adversarial prompts that we re-run on every model update. If a guardrail fails a test, that model version is not released.

6. What you can review

In the Parent App, you can:

• See the full transcript of every session your child has with Lantern
• See the books and worksheets Lantern helped with
• See AI-generated summaries of how each session went
• Adjust safety settings (e.g., enable specific faith discussion) where available

You can also pause sessions at any time, or delete any conversation from the record.

7. Reporting unsafe content

If you observe Lantern produce unsafe, inappropriate, or unexpected content:

• Email safety@lanternlearning.io with the session timestamp and a description of what happened
• We investigate every report within five (5) business days
• If the issue is severe, we may pause the affected model version or capability while we fix it

For COPPA-related concerns, email privacy@lanternlearning.io.

8. Independent safety review

Scout is not currently a member of an FTC-approved COPPA Safe Harbor program. If we join ESRB Privacy Certified, kidSAFE, or another program in the future, that program will include annual third-party audit of our privacy and safety practices.

9. Updates to this page

We update safety guardrails regularly as we learn from real-world use and as AI capabilities change. Material updates to what Lantern will/won't discuss are announced in the Parent App and via email.

Contact

safety@lanternlearning.io — for safety concerns privacy@lanternlearning.io — for privacy concerns 65 Mission Trail Rd, Woodside, CA 94062

Questions? Email privacy@lanternlearning.io. To file a complaint, contact the FTC at reportfraud.ftc.gov or your state attorney general.