An ad captured as a screenshot slips past any text filter — there is not a single word in the message. ModerBot reads the text off the picture and checks it by the same rules as an ordinary message.
Three checks on one image, and each of them can fire independently of the others.
Photos and pictures sent to the chat the usual way, including images with a caption. The picture itself, the text read off it and the caption under it are all checked.
An image sent as a file is checked as a picture. The bot does not unpack the contents of arbitrary documents — PDFs, archives, spreadsheets.
A picture is processed on the fly and is not saved as a file. What exactly stays in the log is described in detail on the security page.
Noticeably longer than text — about a second. That is why an image goes off to be checked separately, while ordinary text messages keep being handled instantly.
Then only the check of the picture itself applies. Heavily blurred, rotated or tiny text is read less reliably — that is a limit of the technology, not a setting.
Individual categories — adult content, ads — can be switched off in the panel. Critical categories stay on whether the violation is in text or in a picture.
Add the bot to your group and grant admin rights — protection starts at once, with sensible defaults.
Add to Telegram