AI Fraud Field Guide
Voice clones cost pennies, deepfakes pass video calls, and scam scripts run at industrial scale. Eight scenarios — all drawn from real cases — test whether your instincts hold. Then keep the playbook.
100% IN YOUR BROWSER · NOTHING ASKED, NOTHING STORED
The Playbook
Five scam families cover nearly all AI-era fraud. Production quality is no longer a signal — the economics and the ask are. Learn the tell; the counter is always a process.
A loved one's voice + panic + an irreversible payment channel.
Family code word. Verify on the number you already have. Real emergencies survive a 60-second callback.
Authority on camera demanding an urgent, unusual transfer.
Out-of-band verification you initiate — policy, not vibes. Production quality proves nothing anymore.
A patient stranger, a private platform, returns that print daily.
The early withdrawal that 'proves' it is the hook. Never invest through a relationship that found you.
Support that contacts you first; any request touching your seed or a signature you don't understand.
Navigate only from your own bookmarks. Seed = never. Read every signature request on the hardware screen.
Look-alike addresses in your history; recruiter code on your machine; links in emails.
Saved contacts + device-screen verification. Untrusted code only in disposable VMs. Type URLs yourself.
The newest reports we tied to this tool. Read one, then come back and use what you learned. 6 reports in the archive.
Educational training scenarios drawn from documented cases. Real scripts evolve — the processes (code words, out-of-band verification, bookmark-only navigation) are what stay reliable. Nothing you do here leaves this page.