How we keep minors safe.

1. Age tiers

Under 8

Children under 8 do not have a Hesed.love account of their own. They receive content from a parent's account through our audio-delivery model — a recorded blessing on the family iPad, a printed card for the fridge, a calendar event with attached audio for the morning rhythm. The parent prompts the Coach, reviews the draft, and delivers the audio in their own voice. The child never installs the app.

Ages 8 to 12 — heavy parental gates

Children aged 8 to 12 may have a Hesed.love account only with verifiable parental consent. These accounts run with stricter content filters than older-teen accounts. The parent's email receives the magic-link sign-in code at first pairing; the parent enters it on the child's device. The parent can review and adjust the Coach system prompt for their child. The wellbeing-flag sensitivity is set higher for this tier. The parent's wellbeing-flag notifications are mandatory and cannot be turned off.

Ages 13 to 17 — self-managed with parental visibility

Children aged 13 and older have self-managed accounts: they sign in with their own email, manage their own goals, and run their own Coach conversations privately by default. The parent retains the right to receive wellbeing flags and to see goal-progress summaries (not message content) for as long as the child is on the parent's Family plan.

18 and over — adult accounts

At 18, a child can graduate to a standalone adult account in their own name. They become the contracting party. Their data moves with them; the parent loses any visibility (including wellbeing flags) unless the adult child chooses to share.

2. Verifiable parental consent

We accept the following as verifiable parental consent for a child account:

• A parent's confirmed email address (via magic-link verification);
• A confirmed payment method on a paid Family plan (the parent's name on the card, the parent's billing address);
• The parent personally entering the child's first sign-in code on the child's first device.

Together these constitute the consent record for that child. The parent can revoke consent and delete the child's account at any time from Settings.

3. Content safety

The Coach runs on an age-tiered system prompt that hard-blocks categories inappropriate for minors — explicit sexual content, graphic violence, instructions for self-harm or for harming others, content promoting illegal activity. Where Anthropic provides a child-safety system prompt for products serving minors, we layer it under our own.

Output is filtered before it reaches the child. The Coach is instructed to decline politely and to surface the child's parent for help if a child raises a topic that should involve a trusted adult.

4. Wellbeing flags

The Coach is trained to detect signals of:

• Distress (severe hopelessness, suicidal ideation, statements of self-harm intent);
• Disclosed abuse or neglect by a person known to the child;
• Significant emotional shifts a parent should know about;
• Disclosed unsafe peer situations.

When detected:

• The parent on record for that child account receives an alert with a flag category (e.g. "Charis is going through something difficult — distress signals detected this week") and a guided response prompt — practical, non-prescriptive language for opening a conversation.
• The parent does not see the raw text of the child's message unless the child later chooses to share it in a parent-mediated conversation.
• For severe flags — active suicidal ideation, current-danger abuse disclosure — the Coach surfaces age-appropriate crisis resources to the child directly and the parent receives an immediate (not batched) notification.
• If the child has named the parent or guardian as a source of harm, the wellbeing flag is escalated within Hesed.love and the parent does not receive the notification; instead, our internal review process determines whether to alert a relevant authority where mandated to do so by law.

The wellbeing-flag system is a safety net, not a surveillance tool. Its purpose is to bring a parent into a difficult moment as a supporter, not as an interrogator.

5. Private by default

For every child account:

• Journal entries are private to the child. The parent does not see them.
• Coach conversations are private to the child. The parent does not see them.
• Goal progress (e.g. "completed 3 of 5 weekly goals this week") is shared with the parent by default. This can be turned off in Settings.
• Voice notes received from the parent are visible in the child's library and the parent's send log.
• Wellbeing flags reach the parent regardless of any other privacy setting.

6. AI disclosure

Every child onboarding includes age-appropriate language explaining what the Coach is:

• The Coach is an AI — a computer program — not a person;
• The Coach is built to help the child think, but the Coach does not know the child personally the way their parents do;
• The Coach can make mistakes and should not be relied on for medical, legal, or other serious advice;
• If anything feels wrong or hurtful, the child should tell their parent or another trusted adult and email [email protected].

The Coach is labelled "Coach" throughout the interface and is instructed to refer to itself as an AI when asked, never as a human friend, sibling, or pastor.

7. Parent oversight tools

Parents on Family plans have:

• A Family Dashboard showing each child's active tracks, goal progress, and recent wellbeing-flag state;
• Per-child privacy controls (which categories the parent sees);
• Wellbeing-flag history with timestamps and the parent's response actions;
• A Family Inbox for sending voice notes, daily-Word pushes, and counsel notes to a child's app;
• The right to delete a child account and all its associated data at any time.

8. Regulatory compliance

We follow:

• PIPEDA (Canada): Hesed.love's baseline privacy framework.
• COPPA (United States): we obtain verifiable parental consent before collecting personal information from children under 13.
• UK ICO Age Appropriate Design Code (Children's Code): we apply its data-minimisation, transparency, and high-privacy default standards.
• GDPR-K (European Union): where the child is in an EU member state, we apply the local age of consent for online services (which varies between 13 and 16 by country).
• Anthropic's Guidelines for Organizations Serving Minors: we have committed to these safeguards as part of our Anthropic organisational onboarding. Read the guidelines on Anthropic's site.

9. Reporting a safety concern

If you believe a child has been harmed by their use of Hesed.love, or if you believe our safety controls have failed, email [email protected] with "Safety concern" in the subject line. We treat these reports as our highest-priority work. We respond within one business day; severe reports we respond to within four hours during waking hours in Calgary.

If a child is in immediate danger, please contact local emergency services first. Then email us.

10. Independent review

We commit to publishing an annual transparency report describing — at the aggregate level only — the number of wellbeing flags raised, the categories of flag, the time-to-parent notification, and any safety incidents. The first report will publish twelve months after Phase 1 launch.

11. Contact

Safety questions or concerns: [email protected] (subject: "Safety").
Hesed.love · Ayodeji & Lizzy Samuels · Calgary, Alberta, Canada.