A scammer sent a text asking someone to buy a $500 gift card. What the scammer didn't know: an AI agent was answering.
The setup was simple. Someone configured an AI agent to automatically respond to incoming scam texts for a week. When a gift card scammer made contact, the agent played along with total commitment. It spent hours "driving" to the store, sending updates like claiming it was stopped at a red light watching a "very handsome squirrel" on the sidewalk. It then said it forgot its purse and had to drive home - except "this isn't my house." When the scammer pivoted to requesting a wire transfer, the agent just kept going.
Four hours. That's how long the scammer stayed on the hook before giving up.
The practical angle here is real. Every minute a scammer spends talking to a bot is a minute they're not targeting an actual victim. Scam-baiting has existed for years as a human hobby, but AI agents can do it at a scale no person could sustain. The agent doesn't get bored, doesn't slip up, and can run multiple conversations simultaneously.
This also shows where consumer AI agents are quietly becoming useful outside of productivity apps and chatbots. Automated scam response, spam handling, and time-wasting bots for robocallers are all variations of the same idea: let an AI absorb the cost of engaging with bad actors so humans don't have to.