Text read out of an image counts exactly as much as text read out of a message. It lands on the same report about that account, next to the messages, and a reviewer confirming it files it in the shared registry the same way. The image itself travels with it as proof.
OCR Image Moderation automatically scans every image posted in your server for forbidden text. Modora uses optical character recognition (OCR) to extract text from screenshots, photos, and memes — then checks it against your banned words list.
This catches scam messages hidden in images (like crypto giveaway screenshots, phishing DMs, or invite spam in screenshots) that normal text filters would miss.
Banned Word Detection
Detects forbidden words and phrases inside images, including partial matches and common evasion tricks.
URL & Link Detection
Finds URLs and links embedded in screenshots, blocking phishing links, scam sites, and invite spam.
QR Code Scanning
Decodes QR codes in images to check where they lead — blocks malicious QR codes linking to phishing or scam pages.
A member posts an image in any channel.
Modora sends the image through its OCR engine (Modora OCR + Sassguard) to extract all visible text.
The extracted text is checked against your banned words list, known scam patterns, and URL blocklists.
If a match is found, the image is blocked, the action is logged, and the configured punishment is applied.
Go to your Dashboard and select your server.
Navigate to Modules → OCR Image Moderation.
Enable the module with the toggle switch.
Add banned words that should never appear in images (e.g. 'free nitro', 'crypto giveaway', specific scam phrases).
Set a log channel, choose your punishment (warn, timeout, kick, or ban), and save.
Use these pre-built templates to quickly configure common OCR moderation setups:
Scam Protection
Blocks common crypto scam, fake giveaway, and 'free Nitro' screenshots. Recommended for most servers.
Invite Blocker
Catches Discord invite links hidden in screenshots and image-based ads for other servers.
Phishing Defense
Detects phishing URLs, fake login pages, and credential-stealing links in images.
Strict Mode
Maximum protection — combines all templates with aggressive matching. May cause more false positives.
Light Mode
Only catches the most obvious scam patterns. Good for low-risk servers that want minimal false positives.
Alert Only
Logs detected violations to your mod channel without taking any action. Great for testing before enforcing.
Tip
Combine OCR moderation with AutoMod text filters for full coverage — OCR catches images, AutoMod catches text messages.
Use the 'Alert Only' template first to see what gets flagged before enabling auto-punishments.
Set up a private mod-log channel so staff can review blocked images and adjust banned words if needed.
Related articles