Bring the thing.
Paste a message, link or selected number. Remove private details and choose source checks or AI interpretation deliberately.
Before you click, pay, or reply.
Take a second look.
A suspicious request. A convincing promise. See what’s being asked, what warrants checking and a practical next step.
Check a message or linkFree beta for adults 18+ in the United States. Capacity limits apply. No safety guarantees.

A payment request. An odd link.
A message that doesn’t add up.
No saved check history in the application. Reports you explicitly send, support correspondence, hosting logs, usage metadata and optional OpenAI processing have separate handling. Clear does not recall a submitted check or report. No advertising trackers. Read the details.
Paste a message, link or selected number. Remove private details and choose source checks or AI interpretation deliberately.
See the request, quoted warning signs and what remains unverified. Source checks and AI interpretation stay separate choices.
Get practical verification steps. Use a number or app you already trust to check a claim independently.
Scams change. Context matters. A real message can look unusual, and a convincing scam can look ordinary. This beta helps you notice clues; it cannot establish that a person, business, or link is legitimate.
Read how the check works“Limited evidence” means we don’t know enough. It never means safe.
Quoted warning signs, source evidence and AI interpretation stay clearly labelled.
No automatic browsing surveillance. No saving your messages in an account history.
It can help you investigate. With your separate lookup choice, OddFlag checks DNS and registration records and, when available, a permitted threat list. It does not open the suspicious site or inspect redirects. Each source has limits; an ordinary-looking link or no list match never means safe.
It checks number structure and, with your lookup choice, a dated public FTC complaint snapshot and a small official-contact directory. Reports are unverified, coverage is limited, and caller ID can be spoofed. A published number does not authenticate the person calling you.
Text and URL patterns work without AI. When available, optional OpenAI review considers your chosen context and available source evidence; web screenshot interpretation also uses OpenAI. AI requires its own per-check choice, can be wrong and has shared capacity limits. Source lookups have a separate choice.
Contact your bank or payment provider promptly using its trusted app or the number on your card. If you shared a password, change it through the real service and review account access. See the FTC’s recovery guidance. This product does not recover money or offer emergency response.
No. Real emergencies, affection and charitable appeals can be legitimate. The check can highlight explicit pressure tied to a request for money, secrets or access. It does not establish anyone’s intent, diagnose a relationship or confirm that a story is true. Subtle tactics and missing context can be missed.
This web beta is free, with request and shared AI capacity limits. No subscription or card is offered here. Features can be temporarily unavailable; no unlimited allowance is promised.
The web checker is the reviewable beta. Mobile, widget, and browser integration source is being prepared separately. No app-store approval, public listing, or automatic protection is claimed.