When a member thinks they were punished by mistake, they can appeal. Open Appeals in the group dashboard, read the reason, and click Approve (lifts the punishment and teaches the bot it was a false positive) or Reject (keeps the punishment). Many clear cases are decided automatically by an AI pre-check before they ever reach your queue.
Want to set this up? Do it right in your Telm account.
Open in your dashboard1What an appeal is
An appeal is a member's request to review a moderation action against them. When someone is banned, muted or has a message removed, the bot offers them a way to contest it: a button under the notice in the group, a deep link that opens the bot in a private chat, or the /appeal command.
Each appeal carries the original decision it disputes, so you are never judging blind. You see who the person is, when they appealed and the reason they typed, and you can open the underlying decision to see exactly why the bot acted.
2Where to review appeals
Open the group in your dashboard and go to the Appeals page. It lists every pending appeal for that group, newest first, with a small summary: how many are pending and how many were already resolved automatically.
Each row shows the appeal number, the member (username, name or numeric id), the reason they gave and the time it arrived. Approve and Reject buttons sit on the right of every row.
3How to decide
Read the member's reason first, then weigh it against the original decision. If the removal or ban was a genuine mistake, approve. If the content really was spam or broke your rules, reject.
4The AI pre-check
Before an appeal reaches your queue, Telm can decide the obvious cases for you. An independent AI arbiter re-reads the punished message with full context, with strict anti-abuse limits so it runs at most once per incident.
If the pre-check is confident the message was clean, the appeal is auto-approved and the member is unbanned without you lifting a finger. If it is confident the message really was spam, the appeal is auto-rejected. Everything in between is left for a human, so your queue only holds the genuinely ambiguous cases. The summary at the top of the page shows what share of appeals were settled automatically.
Appeals against actions a human moderator took by hand are never auto-decided against you: the AI verdict is treated conservatively, so it can only ever suggest leniency, not overturn a manual ban.
5Reviewing from the bot
You do not have to open the dashboard. Any admin can manage appeals right in the group chat.
- /appeals — lists the open appeals for this group with their numbers.
- /approve <id> — approves appeal number <id> (unban plus false-positive learning).
- /reject <id> — rejects appeal number <id> and keeps the punishment.