2026-05-14

When ad bots invade your chatroom, keep auto-reply from talking to them

When ad bots invade your chatroom, keep auto-reply from talking to them

"An ad bot joined the chatroom and started spamming, and our auto-reply started answering it. Two bots having a conversation in our chatroom."

Common failure. Auto-reply does not distinguish senders, it reacts to any new message. So when an ad bot arrives, your operator-toned replies start answering it.

New message → reply decision flow

Replyer's persona runs every new message through 3 sequential filters:

New message @_bot_user · "buy now https://..." Filter 1 Trigger pattern Filter 2 Keyword block Filter 3 Sender ban Reply No reply (ad bot blocked) Input 5+ chars target lang / no URL / username "ad", "event", t.me/+, bit.ly Telegram Block User

Three separation strategies

1. Trigger patterns (persona-side defense)

  • "Body must contain 5+ target-language characters" - filters off-language ad messages
  • "No URLs" - ad bots almost always include URLs
  • "Sender username does not match known bot patterns" - exclude _bot / _official

2. Keyword blocks (chatroom-side defense)

  • "ad", "promotion", "event participation" - obvious marketing vocabulary
  • "admin will guide you to a temporary channel" - classic Telegram scam bot
  • URL domains (e.g. bit.ly, t.me/+)

3. Sender bans (Telegram-side defense)

  • Block only applies to the operator's view
  • Removing the bot from the chatroom needs admin privileges
  • Bots re-join under new accounts often

Ad bot vs human - 6 signal scores

Probability each signal points to ad bot vs human. Heavy left = automatic filter target.

Ad botHuman member
Username pattern
No profile photo
Repeated phrasing
Contains URL
Speaks on join
5+ msgs/minute

Recovery flow (5 steps)

If the accident has already happened:

1
Auto-reply OFF immediately

Hit [Stop] or trigger persona vacation. Stops further pile-ups.

2
Operator announcement in chatroom

"Ad bot joined and pulled auto-reply into a conversation. Paused and tightening block rules." One sentence preserves trust.

3
Tighten triggers / keyword blocks

Extract common keywords / URLs / patterns from bot messages → update persona rules.

4
Clean response history noise

Persona → response history → bulk delete / mark [inappropriate] bot conversations.

5
Resume auto-reply

24h manual ops, then 1-2 days [manual review] mode, then auto.

Operational pattern for chatrooms with frequent ad-bot invasions

  • Operator approval required for new members (admin permission)
  • Exclude new members from auto-reply for first 48 hours
  • Distribute admin permissions to 2-3 trusted members
  • Weekly review of ad-bot patterns + block-rule updates

FAQ

Q. Don't aggressive ad-bot filters block real members too?

Possible - false positives. Mitigation: use [manual review] mode - filtered messages land in the queue for the operator to fire manually.

Q. What if an impersonation bot copies our operator tone?

  • Pin operator's Telegram username in the chatroom intro
  • Use a consistent marker in operator auto-replies
  • On discovery, announce in chatroom and tell members how to verify

Q. Should block rules differ per chatroom?

Yes. Tighter for info chatrooms, looser for social rooms. Replyer's persona × chatroom matrix supports per-room mapping.

Q. What if the ad bot scrapes operator messages to use elsewhere?

Scraping risk. Tighten the persona so auto-replies don't include marketable content. Access control on the chatroom helps too.

Q. After tightening rules, sometimes the operator bot stops responding to humans.

Rules got too strict. Check Diagnostics "no-reply rate" → review filtered messages → loosen rules.

Next step

Grab the build for your OS from the Replyer download page and follow the usage manual for step-by-step setup.