Image Spam in Telegram: The Threat Your Keyword Bot Can't See

Open your Telegram group right now. Scroll through the last week of spam your bot caught. Now ask yourself: how many of those were images?

If the answer is "none" — it doesn't mean your group doesn't get image spam. It means your bot can't see it.

The shift to visual spam

Spammers are smart. They watch what bots detect and adapt. When keyword filters became common, they switched to Unicode tricks and Cyrillic lookalikes. When AI text analysis got better, they shifted to images.

The logic is simple: if the spam is in a JPG, your text-based bot can't read it. A screenshot of a "guaranteed profit" offer bypasses every keyword filter, every regex pattern, every text-based AI model. The bot doesn't even know there's text in the image.

In 2026, image spam takes several forms:

Screenshots of offers. A casino ad, a "crypto trading course," a fake job listing — rendered as an image instead of text. Everything that would trigger a text filter is invisible to the bot because it's pixels, not characters.

QR codes. A QR code posted in a group can link to anything — a phishing site, a malware download, a fake wallet address. To the bot, it's just a square pattern in an image. To a curious member who scans it, it's a trap.

Edited screenshots. A screenshot of a fake conversation — "Look, I made $5,000 with this method!" — designed to build credibility for a scam. Text-based moderation can't detect fabricated conversations embedded in images.

Sticker and media spam. Forwarded stickers, GIFs, and short videos with embedded promotional content. Some spammers create custom sticker packs specifically for spamming groups.

Why most bots ignore images

Processing images is expensive. Text analysis is fast and cheap — you parse a string, run it through filters or an AI model, done in milliseconds. Image analysis requires computer vision: OCR (optical character recognition) to extract text, object detection to identify what's in the image, QR code parsing, and then analysis of the extracted content.

Most Telegram bots skip this entirely. Combot, Rose, Shieldy — none of them analyze image content. Some AI bots do basic image analysis but not QR code detection.

The result: a growing blind spot. Spammers know which bots are popular and what they can't see. Image spam is increasing specifically because it works.

What image-aware moderation looks like

A bot that actually handles image spam does several things:

It extracts text from images using OCR — reading the words embedded in screenshots, photos, and edited images. It then runs that extracted text through the same spam detection pipeline as regular messages.

It detects and decodes QR codes, then checks where they link. A QR code pointing to a known phishing domain gets the same treatment as a phishing link posted as text.

It analyzes image content beyond text — recognizing casino imagery, fake trading dashboards, adult content, and other visual patterns associated with spam.

It considers context. An image of a product in a marketplace group is normal. The same image in a tech support group might be spam. Group-specific understanding matters for images just as much as for text.

Varta does all of this. When someone posts an image in a Varta-protected group, it doesn't just see "a photo was shared." It reads the text inside, checks any QR codes or links, evaluates the visual content, and makes a judgment — same as it would for a text message.

The admin's blind spot

Here's what makes image spam particularly dangerous: admins often don't realize it's happening.

With text spam, you see it even if the bot misses it — the message is right there in your chat. With image spam, most admins scroll past images quickly, especially in active groups. A casino ad screenshot looks like just another photo until you actually look at it. By then, 50 members may have already seen it.

And because your bot doesn't flag it, there's no notification. No "spam detected" alert. No record in the moderation log. It's invisible to both the bot and the admin.

The first step is awareness. Check your group for images posted by recently joined accounts. If you see promotional screenshots, QR codes, or suspiciously professional graphics — you have an image spam problem. Your bot just hasn't told you about it.

The second step is switching to a bot that can actually see.

Add Varta → image and QR analysis included in every plan