Every action Telm takes is logged with a reason. Open the group journal in your dashboard to see exactly which rule fired and why. If it was a genuine mistake, the punished member can appeal in a couple of taps and an AI arbiter reviews it fast.
1Start with the journal
The moderation journal is your source of truth. Every removal, mute and ban is recorded there with the message, the member, and the reason it was flagged.
You never have to guess why something happened — the decision detail spells out which detector or rule matched.
- Open the group in your dashboard and go to the journal or decisions view.
- Find the entry for the message or member in question.
- Open the decision detail to see the matched rule, the confidence and the action taken.
2Common reasons for an action
Most actions come down to one of a handful of causes. Knowing them helps you read the journal quickly:
- Spam or scam content — the AI or a filter recognised known spam, a scam pattern, or a contact-out attempt.
- CAS or global ban — the sender is a known spammer in a global ban database and is blocked on entry.
- A custom rule — one of your own if-then rules matched the text, link or media.
- New-member screening — a first-time poster tripped a stricter check applied to newcomers.
- A banned link or word — the message contained something on your block lists.
3If it looks like a mistake
False positives are rare, but they happen. If the journal shows an action you disagree with, you have two quick paths.
- As an admin, you can reverse the action directly — unban or unmute the member and, if needed, add a rule or whitelist entry so it does not recur.
- Report the decision as a false positive from the journal so the system learns and improves across all your groups.
4How the punished member appeals
When the bot punishes someone, it sends that person a notification in their own language with an appeal button (if they have messaged the bot before; otherwise the notice appears in the group). They do not need an account with Telm to use it.
The appeal is reviewed by an independent AI arbiter on a stronger model, and a clearly wrong action can be reversed within minutes.