Extract the request
AI can help structure the exact action in the message, but the quote must stay grounded to the artifact.
Before you call, click, pay, or reply.
Check what a court message is asking you to do against independent public sources.
No independent source, no match. If evidence cannot establish a claim, SEAL leaves it unconfirmed.
The City of Dallas published this court-looking QR-payment notice as a scam example. SEAL checks the requested action against that independent source.

“Scan the QR code to settle your unpaid balance.”
The City of Dallas published this exact notice as a scam warning. That is enough to reject the QR-payment route without trusting anything inside the message.
The model can structure what the message asks. It cannot decide the verdict. SEAL checks outside the message, then applies a fixed evidence boundary.
AI can help structure the exact action in the message, but the quote must stay grounded to the artifact.
Reviewed court and government resolvers compare claims against sources the sender does not control.
A match or conflict requires cited evidence. Without it, the result stays unconfirmed.
A real court name, address, phone number, or public case record can appear inside a fraudulent message. Matching details never authenticate the sender or requested action, and model output cannot promote a claim to a verdict.
Check it before you act.
Check a court message