⚖️ Blify Mod

Privacy and data

Blify Mod is, by necessity, a service that reads chat messages. This page is the plain-English version of what gets sent where, what's stored, and for how long. The legal version lives at /privacy.

#What gets sent to AI providers

When the bot scans a message, the message content is sent to:

  • OpenAI (gpt-5-nano) — used as a fast pre-screen on every tier. If OpenAI is unavailable, the screen falls back to Claude Haiku 4.5.
  • OpenAI (gpt-5-mini) — used for the actual moderation decision on the Free tier. If OpenAI is unavailable, the judge falls back to Claude Haiku 4.5.
  • Anthropic (Claude Sonnet 4.6) — used for the actual moderation decision on Pro/Premium tiers. If Anthropic is unavailable, the judge falls back to OpenAI gpt-5-mini.

Both providers are sent only the content being scanned and a small slice of channel context needed to make the decision (typically 8–20 messages). Messages from non-moderated channels are never sent anywhere.

Important: Anthropic does not use Blify Mod's API traffic to train their models. Blify Mod has not opted into any training program with either provider.

#What we store ourselves

Blify Mod uses a MongoDB database. Per server, we store:

  • The bot's configuration: enabled/disabled, moderated channel IDs, log channel ID, custom rules, immune roles/users, strictness, dry-run flag, appeal link.
  • A moderation history for each user who has been actioned: action type, severity, rule, reason, channel ID, the first 500 characters of the offending message, the timestamp, and whether it was reversed.
  • A rolling per-day total of AI API cost incurred for this server (in USD), used to enforce the per-tier daily moderation throughput cap. These records contain no message content — only the guild ID, the date, and the running cost total. Auto-purged after 30 days.

Authorized members of the Blify Mod operations team can access this stored configuration (including your custom rules) to investigate abuse reports or rule setups that may break Discord's Terms of Service or Community Guidelines.

Per user (across all servers), we store:

  • Vote credits, daily streak, and last-vote timestamp (only for users who have ever voted).
  • Active Pro/Premium role redemptions, with their expiry timestamps.
  • Pro/Premium claim records (which servers your subscription is currently active on).

Web sessions for the dashboard are stored in MongoDB with a 7-day TTL.

We do not store:

  • Message content from non-moderated channels.
  • Message content from moderated channels that was scanned and found clean.
  • Anything from DMs.

#Retention

  • Server config is kept while the bot is in your server. When the bot leaves a server, all of that server's config and moderation history is deleted automatically.
  • User credits and Pro records are kept for as long as the account exists.
  • Web sessions expire after 7 days of inactivity.

#Telling your members

Discord requires bots that send messages to AI services to disclose this to users. We recommend pinning a note like:

This server uses Blify Mod for AI moderation. Messages in #general, #off-topic, and #help may be sent to OpenAI's and Anthropic's APIs for evaluation. Neither provider trains on this data. See https://blifymod.com/privacy for details.

The bot's welcome DM (sent to whoever invites it) includes a similar note you can copy.

#Deleting your data

  • Server data: kick the bot. Everything for that server is deleted within seconds of the GuildDelete event firing.
  • Personal data (credits, Pro records, dashboard sessions): open a ticket in the support server — owner deletes within 7 days.
  • Moderation history about you, in someone else's server: that's the server owner's data, not yours, since they configured the bot to log it. Ask the server's staff.