The inbox is the front line. Now the attacker writes like you.
Generative AI has erased the old phishing tells. No typos, no broken grammar, no clumsy translations — just flawless, personalised messages, cloned writing styles, deepfaked voice notes and hijacked real threads. This is a working simulation: you'll sit in a real inbox during a high-pressure hour and decide what to trust.
You are Dana Okafor, Senior Treasury Analyst at Meridian Atlas Group. It's 15:10 on quarter-end Friday. The 16:00 payment run is locked once you approve it. Eight messages are waiting — vendors, your CEO, IT, HR, the CFO. Some are routine. Some were written by an adversary with an AI model and a copy of last month's email thread.
For each message you can inspect the sender and headers, hover a link to see where it really goes, or verify out-of-band before you act. Then decide: report it, keep it, or — when the message asks you to — comply. Catch the threats. Don't cry wolf on the real mail. And whatever you do, don't approve a payment to an account you didn't verify.
Have an idea for this game?
Every Insights game takes submissions — ideas, feature requests, and enhancements — straight from the practitioners who play it.