QR code moderation. Detect QR codes in images

Content moves through four layers. ML detection finds and decodes every QR code in the image and extracts its URL. Custom rules match that URL against the allow, block, and route lists your team controls. The AI Moderator weighs grey-area URLs against account and content signals, and the small remainder reaches a human reviewer with the image, the URL, and the AI Moderator's notes. Roughly 99.9% of decisions are automated, and recurring patterns become new rules.

Detection holds up on codes placed in corners, covered by stickers or watermarks, and on several codes in one frame. The layered pipeline automates around 99.9% of decisions and routes the genuinely unclear cases to human review, so a borderline read is checked rather than guessed. Specific detection-rate figures should be confirmed against current product data before publishing.

Text URL filters only scan the text a user types, so a link encoded in an image slips past them entirely. QR code moderation reads the image itself, decodes the embedded URL, and runs it through the same domain rules you apply elsewhere. That closes the gap users exploit to push contacts, payments, and phishing links off-platform.

QR codes encode URLs rather than language, so detection and URL extraction work the same regardless of the surrounding content's language, and the rules engine runs across Lasso's 200+ supported languages for any accompanying text. It runs through the API and dashboard as part of the four-layer pipeline, so it scales with the rest of your moderation rather than as a separate tool.

Four steps. First, every QR code and its URL are detected. Second, the URL is checked against your rules. Match an allowed domain, it passes. Match a blocked domain, it is rejected or flagged. Edge cases pass to layer three, the AI Moderator. If the AI Moderator is unsure, the QR is passed to layer four, the human moderator.

Lasso extracts the URL and applies your rules. It does not follow the URL to scan the destination. This is deliberate. Your rules are predictable and under your control. Third-party destination categorization can change silently when a vendor updates a model.

Rules are built and edited in the dashboard by your team in minutes. Add URLs or full domains, set the action for each, and they apply automatically. Change a rule, and every subsequent QR code is checked against the new rule.

Yes. For video, see the video moderation feature, which applies the same QR detection and the same URL rules to video frames.

Three verticals get the most value today. Dating: prevent profile-photo QR codes that move conversations off-platform. Marketplaces: prevent listing images from routing buyers to competitor sites. Social: block phishing and spam QR codes that bypass text-based URL filters.