Setting Boundaries: How Admin Controls Keep Family AI Safe

When you give your family access to an AI assistant, someone needs to be in charge. Not in a controlling way — in the way that someone manages the Wi-Fi password, the Netflix account, or the family phone plan. Administrative control over a family AI isn't about surveillance. It's about structure.

What Admin Controls Should Include

A family AI assistant without admin controls is just a shared chatbot. Here's what real admin controls look like:

Member Management

The admin (typically a parent) should be able to:

• Add members via invite code — no need to share passwords or account credentials
• Remove members when someone leaves the household or access needs to change
• View the member list with roles clearly displayed (admin, member, minor)
• Transfer admin status if household dynamics change

With Ori, creating a family group generates an 8-character invite code. Share it with family members, and they join the group instantly. The admin can remove anyone at any time.

Minor Designation

This is the most critical admin control. The ability to designate a family member as a minor changes the AI's behavior for that specific user:

• Content safety classification activates — every message is evaluated before the AI responds
• Tool restrictions engage — web search, webpage reading, and third-party integrations are blocked by default
• Age-appropriate system prompt applies — the AI becomes educational, encouraging, and safety-conscious
• Monitoring begins — flagged messages are surfaced to the admin

The admin configures these settings per minor. A 16-year-old might have different restrictions than an 8-year-old. The settings are flexible, not one-size-fits-all.

Safety Dashboard

The admin dashboard is where flagged content lives. For each flag, the admin sees:

• Who sent it — which family member's message was flagged
• The message content — truncated for quick scanning, expandable for full context
• The category — what type of concern was detected (explicit, violence, drugs, self-harm, etc.)
• The severity level — low, medium, or high
• Ori's response — how the AI handled the situation
• Review status — whether the admin has acknowledged the flag

This dashboard isn't a transcript of every conversation. It's a focused view of only the messages that the safety system identified as concerning. The admin reviews flagged items, marks them as reviewed, and determines if a conversation with the child is needed.

Credit Management

A practical but important control: the admin manages the group's AI usage credits. Family members can use personal credits or group credits for their conversations. The admin chooses the plan, manages billing, and allocates resources.

This prevents the scenario where one family member exhausts the entire AI budget on a single evening of creative writing.

The Balance Between Control and Privacy

Good admin controls respect privacy. Here's what the admin can and cannot see in Ori:

The admin CAN:

• See flagged messages from designated minors
• View group chat messages (visible to all group members)
• Manage member roles, settings, and permissions
• Configure sharing permissions for group knowledge

The admin CANNOT:

• Read private conversations between any member and Ori
• See unflagged messages from minors
• Access another adult member's personal data or chat history

This boundary is essential. A family AI that lets parents read every message isn't a family tool — it's a surveillance tool. Admin controls should provide oversight where it's needed (child safety) without overreaching into areas where privacy should be preserved (adult conversations, unflagged minor conversations).

Why This Matters

As AI becomes a daily utility — like electricity or internet access — the concept of "family AI admin" will become as natural as "family phone plan administrator." Someone needs to manage access, set appropriate boundaries, and ensure the technology serves the whole family safely.

The families who set this up now, while their children are still forming their relationship with AI, will be ahead of the curve. Not because the technology is complex, but because the habits and boundaries they establish early will shape how their children interact with AI for years to come.

Ori is built for these families. One admin. One group. Clear controls. Real safety. And the flexibility to evolve as your family grows.