Spam protection

How the detection engine works and how to tune it for your community.

How Telm spam detection works: layers and verdictsA plain-language tour of how Telm stops spam: several layers of protection, our own ML model, AI for the tricky cases, and controls that keep you in charge.

AI moderation in Telm: the LLM classifier explainedTurn on AI moderation to catch spam that hides its intent: a language model reads meaning, routes doubtful cases to review, and you enable it per group.

The CAS blocklist in Telm: shared spammer protectionTurn on the CAS (Combot Anti-Spam) integration so known Telegram spammers are blocked at the door, before their first message. Available on every plan.

Global spam database and federated bansHow Telm detects coordinated spam across all protected groups and bans a confirmed spammer everywhere at once, with safeguards and full reversibility.

The personal ML model in Telm: trained on your chatUpload a Telegram export and Telm trains a personal ML layer that vetoes false spam alarms on your community's normal speech, with strict quality gates.

Voice moderation in Telm: audio checked like textTranscribe voice messages and round video notes and run the text through the same Telm anti-spam pipeline, so spam cannot hide in audio. Limits explained.

Content Filters — Block Links, Media, Words and MoreThe complete guide to Telm content filters: link allow and block lists, external promotion, forwards, images, stickers, GIFs, video, profanity and stop-words.

Language Policy — Enforce Your Group's LanguageDeclare which languages belong in your Telegram group. A message in a script you have not allowed is deleted or the sender banned. Detection is local.

Custom Moderation Rules — Your Keywords and PatternsCreate your own moderation rules from keywords and patterns, beyond the built-in engine. Match on words, regular expressions, links and media.

Triggers — No-Code If-Then Automation for Your GroupBuild if-then automations with a click: match a condition, run an action — auto-replies, auto-deletes, warnings, mutes, kicks and escalations, with no code.

Moderation Journal — Every Decision the Bot MakesSee who was moderated, when, why and by which layer. Search, statistics and CSV export, plus label detections as spam or false positive to train protection.

Whitelist and Member Trust — Exempt Trusted PeopleWhitelist trusted members and domains so your regulars are never moderated, let reputation soften checks, relax rules per group, and use whitelist-only mode.

How to Create a Custom Moderation Rule in TelmStep-by-step tutorial: build your own keyword or pattern-based moderation rule in Telm, pick an action, set priority and go live in minutes.

How to Set Up a Trigger (Auto-Reaction) in TelmStep-by-step tutorial: create a trigger in Telm that auto-replies or moderates when a message matches a keyword, mention, command or regex pattern.

How to Whitelist Allowed Domains for Links in TelmStep-by-step tutorial: turn on the link filter in Telm and whitelist the domains you allow, so only trusted links pass and spam links are blocked.

OCR image spam detection in Telm: text inside picturesHow Telm reads text inside images, recognises known spam banners and decodes QR codes so image-only spam is caught as fast as plain text.

Moderation strictness levels in TelmLight, Medium and Strict moderation presets in Telm: exactly what each level turns on, who they suit, and how to fine-tune after choosing.

Controlling forwards and channel promotionStop third-party promotion: block forwards, channel posts, messages sent via bots and links to outside channels, with a whitelist for the ones you trust.

How Telm detects scams and phishing in Telegram groupsHow Telm catches crypto scams, phishing, fake giveaways and investment fraud in Telegram groups by scoring dozens of signals and reading intent, not keywords.