How it works
Check Trust Score app reads your screenshot with a multimodal AI model trained on language, design, and patterns common to phishing, smishing, and impersonation scams.
The 3-step analysis
- 1. Extract. We pull the sender, body text, links, and visible branding out of the image.
- 2. Evaluate. The model checks for known fraud patterns: urgency, lookalike domains, requests for payment/credentials, mismatched sender info, and brand impersonation.
- 3. Score. You get a 0–100 risk score with a clear verdict, the specific red flags found, and a recommended next step.
What the score means
- 0 – 30 · Likely Safe — No strong fraud indicators. Still verify if money or credentials are involved.
- 31 – 69 · Suspicious — Some warning signs. Don’t click links; contact the organization directly.
- 70 – 100 · Likely Scam — Strong fraud indicators. Do not respond, click, or pay.
Limits to know
- AI judgment is probabilistic, not definitive. Treat the score as guidance.
- Screenshots can be doctored or partial — context matters.
- We don’t (yet) cross-check links against live threat databases. A clean-looking URL can still be malicious.
- When real money is at stake, always call the organization on a number from their official website — never a number from the message.
Privacy
Your screenshot is analyzed in memory and discarded after the response. No account, no history, no tracking of message contents.