AI interaction transparency
This page explains how Salty Panda discloses that Panda is AI, how crisis-resource referrals are tracked in aggregate, and how our current retention language is meant to work in plain English.
What this page is for
This report is intentionally redacted. It does not publish conversation content, memory text, or private user data. It only shows the current disclosure posture, safety protocol version, and aggregate companion-safety reporting.
1. Current companion posture
- Audience
- Adults 18+ only.
- Chat disclosure
- Panda is AI, not human, and can make mistakes.
- Channel disclosure
- Identity is disclosed at the start of paired channels and re-shown during long-running conversations.
- Public safety page
- Open the Safety Protocol
2. Safeguards we keep live
- Panda discloses that it is AI and not a human being.
- Crisis language is routed to crisis resources instead of normal companion behavior.
- Paired channels use reminder cadence instead of repeating a heavy warning in every message.
3. Retention language in plain English
- Disconnect
- Disconnect stops future sync and new provider access.
- Purge
- Purge removes app-side data only where that provider or surface supports it.
- Account deletion
- Account deletion removes active app-side personal data and starts downstream cleanup for supported providers.
- Backups
- Encrypted backups may retain data briefly until normal backup rotation ages them out.
4. Reporting notes
The aggregate counts on this page are generated from our companion-safety reporting path. They are meant to show whether crisis-resource routing is happening and which channels are involved, without exposing user content.
If you need the companion safety protocol itself, read the Safety Protocol. If you need legal/privacy details, use the Privacy Policy and Terms of Service.