GroupButler is an AI system and never presents itself as a human. Automated actions (mute/ban/delete) always carry a machine-readable reason code and are logged for admin review. Low-confidence decisions route to a human-review queue, not automatic punishment.
Token risk screening uses aggregated public data from GoPlus, RugCheck and CertiK APIs — we send only the contract address, never message content. LLM inference runs on contract-bound providers; message text sent for analysis is not retained by us beyond verdict metadata.
Verdict metadata: 90 days rolling. Drill/test data: published publicly only in aggregate form with no user-identifying content. On request, we delete all data associated with your user ID or group within 30 days: contact [email protected] or remove the bot from the group.
Access, correction, deletion, objection — email [email protected]. We respond within 30 days. GDPR/CCPGA-style rights honored for all users regardless of region.
The service is not directed at children under 13; we do not knowingly process their data.
Material changes will be announced in our channel (t.me/web3groupguardian) 7 days before taking effect.