Telm is a Telegram anti-spam and moderation bot. It has a Free plan, it does not read private chats, it works in many languages, and it checks messages in real time without storing them permanently. The details are below.
1What do I get to start?
There is a Free plan for groups up to 1,000 members, and one account can keep up to 5 groups on it. Each group gets the core anti-spam engine — spam detection and rules, AI moderation, ban appeals, the global CAS ban list, the global spam database and image checks — so you can get real protection running in minutes.
Pro adds voice moderation, a personal ML model, DM broadcasts, deeper analytics, higher member and rule limits and the API; Business covers five groups on one subscription with no member limit. You can try Pro for 7 days without a card to see the difference before you decide.
2Is it private? Does it read our messages?
Telm only processes messages in groups where it is an admin, and only to moderate them. It never reads members' private one-to-one chats — that is not technically possible for a group bot.
Messages are checked in real time; what is kept afterwards — message history, moderation records and analytics — is retained for the analytics window of your plan.
- Sign-in is through Telegram — there are no passwords or emails for Telm to store.
- White-label bot tokens are encrypted and sessions use secure cookies.
- Your data is never sold or shared with third parties.
3What languages does it support?
The website and dashboard are available in many languages, and punishment notifications reach members in the language of their own message.
The detection model itself is multilingual — it is trained on spam across scripts, with special handling for CJK text and for tricks that mix alphabets to dodge filters.
4Will it ban innocent people?
This is the most common worry, and Telm is built around it. Three layers protect real members:
- Monitoring Mode — every group starts in observe-only mode, so the bot punishes no one until you have seen it is accurate. See Monitoring Mode explained.
- Mute-first and reputation — a first mistake usually means a removed message, not a ban, and trusted long-time members are barely checked.
- Appeals (every plan) — if a mistake does happen, the person gets a polite notification and can appeal, reviewed by an AI arbiter. See How appeals work.
5How do I get started?
Setup takes a couple of minutes and needs no tokens, configs or command sheets. You sign in on the website with Telegram, add the bot to your group as an admin, and finish a short wizard.
The bot then starts in Monitoring Mode so you can watch it before it acts.