Telegram Anti-Spam Bot: Two Free Layers That Stop Spam Automatically
A Telegram anti-spam bot is an automated account that reads every message and every join in your group and removes spam before members see it, without an admin watching the chat. Telegram Bot App does this with two independent anti-spam layers that both run free: AI Spam Intelligence, which scores the account behind each message, and Spamfinder, which scores the content of the message itself. A CAPTCHA check at the join stops automated accounts getting in at all, and absolute rules block the invite links and forwards that spam relies on. This article explains what an anti-spam bot is, how the layers work together, and how to switch it on. For the deep configuration and threshold-tuning walkthrough, see the anti-spam configuration guide.
Why one filter is not enough
Spammers adapt to whatever single method a group uses. A keyword blocklist fails when they rephrase. A link blacklist is useless against fresh domains. Simple rate limiting misses bots that post at human speed. An anti-spam bot worth using runs several independent checks, so that a message slipping past one still meets the next. Telegram Bot App pairs a content check with a behavioural check, then adds a join gate and absolute content rules on top.
Layer one: AI Spam Intelligence (behavioural)
AI Spam Intelligence evaluates the user, not just the message. Every account receives a spam rating between 0.0 (trustworthy) and 1.0 (definite spammer), calculated from violation history across groups, account characteristics, and behavioural patterns such as how quickly the account posts after joining. The rating updates continuously as the system observes behaviour, and it cross-references external spam databases, so an account flagged elsewhere arrives already known.
When a user's rating reaches or exceeds a fixed 0.75 threshold, the bot removes them automatically on their next message or join event. Because it acts on the account rather than the content, it can remove a known spam account the moment it joins, before it posts anything. This is a free-tier feature. The full behaviour is described in the AI Spam Intelligence guide.
Layer two: Spamfinder (content)
Spamfinder is a machine-learning classifier that analyses each message and produces a spam confidence score between 0.0 and 1.0. It examines language patterns typical of promotional content, structural signals such as excessive punctuation or all-caps, link density, and known spam phrase fingerprints. You set the threshold that triggers enforcement with a slider from 0 to 100%; the default is 75%, which catches obvious spam while keeping false positives low. Messages over the threshold are deleted and the sender is restricted. Spamfinder is also free, and its calibration is covered in the spam pattern detection guide.
The two layers are complementary. Spamfinder catches content-based spam even from a new account with no history; AI Spam Intelligence catches behavioural spam even when the wording is clean. A campaign has to defeat both at once.
The join gate: CAPTCHA
Most spam plans depend on joining and posting before anyone reacts. A CAPTCHA challenge at the join breaks that plan. Every new member is kept read-only until they solve a simple task, with a timeout you set from 1 to 60 minutes (15 is the default) and a choice to remove or restrict accounts that do not verify in time. A script joining hundreds of groups cannot solve an individual challenge for each, so automated joining stops at the door. Setup is in the CAPTCHA verification guide.
Absolute rules: forwards, invite links, and media
Some spam tactics are best blocked outright rather than scored. From Basic Protection you can block forwarded messages, block Telegram invite links (t.me/... and @channel patterns, which also catches WhatsApp group links), and restrict specific media types. Malicious URLs are detected through Google Safe Browsing and fed into the moderation decision. These absolute rules stop entire categories of spam regardless of how the content is phrased. The prohibited-content controls are detailed in the prohibited content guide.
What happens to a spammer
When any layer flags a violation, the decision engine applies a timed restriction proportional to the severity, roughly 5 minutes for spam or invite links, and escalates the duration for repeat offenders based on their cumulative history. The restriction is a timed mute: the account can read but not post, and is unrestricted automatically when the time elapses. Because posting more is what piles up penalties, a spam account that relies on volume removes itself quickly, while a legitimate member who trips a filter once recovers fast. The model is described in the automated punishment system guide.
How to set up the anti-spam bot
- Add the bot to your group and grant it admin rights, following the step-by-step guide.
- Enable both spam layers from Settings > AI Moderation: turn on AI Spam Intelligence and Spamfinder. Leave Spamfinder at the 75% default to start.
- Turn on CAPTCHA so automated accounts are stopped at the join.
- Enable invite-link blocking under Basic Protection, and block forwards if your group sees forward spam.
- Watch the live punishment feed for a few days, then adjust the Spamfinder threshold using the threshold optimization guide if you see missed spam or false positives.
Both spam layers are free, so a working anti-spam setup costs nothing. For a broader picture of stopping spam bots specifically, see how to stop spam bots on Telegram.
Frequently asked questions
What is a Telegram anti-spam bot?
A Telegram anti-spam bot is an automated account that reads every message and join in a group and removes spam automatically. Telegram Bot App uses two free layers: AI Spam Intelligence scores each account from 0.0 to 1.0 and auto-kicks at 0.75, and Spamfinder scores message content and deletes anything over a threshold you set. A CAPTCHA check at the join and absolute rules against invite links and forwards round out the protection.
Is there a free Telegram anti-spam bot?
Yes. Both of Telegram Bot App's anti-spam layers, AI Spam Intelligence and Spamfinder, are free for every group, along with CAPTCHA, invite-link blocking, and custom badwords. The free tier also includes 500 image scans and 1,000 sentiment checks per month. Paid plans add scan capacity and extra media moderation, but core anti-spam is free.
Does the anti-spam bot remove spammers before they post?
It can. AI Spam Intelligence evaluates the account rather than the message, so an account whose spam rating already exceeds 0.75, for example because it was flagged in other groups, is removed on join before it posts. Spamfinder then handles spam from accounts that arrive without a history, by scoring the content of what they send.
Will it remove legitimate members by mistake?
Rarely, and you control it. The 0.75 auto-kick threshold requires substantial negative indicators, so an ordinary member who trips a rule once keeps a low rating. Spamfinder's sensitivity is a threshold you set, and the reputation system treats established members more leniently than brand-new accounts. Every automatic action is logged, and you can open a user's intelligence report to see why they were flagged and reverse a false positive.
How is this different from the anti-spam configuration guide?
This article is an overview of what the anti-spam bot is and how to switch it on. The anti-spam configuration guide goes deeper into threshold strategies, recommended settings per community type, handling false positives, and responding to active spam campaigns.