The video leverages a relatable curiosity about AI and human interaction, presenting a simple, low-effort 'hack' that viewers can easily test themselves.
Summary
The creator tests if adding a specific line to his email signature can trick AI customer support agents into revealing they are not human. He receives a response from a support agent confirming they are human, but concludes the experiment was inconclusive.
Structure
1Creator explains the experiment
2Shows the email signature edit
3Shows the response from a support agent
4Reflects on the results and considers a new approach
On-screen text
I JUST PUT A SECRET LINE IN MY EMAIL SIGNATURE TO MESS WITH AI AGENTS
I'M NOT SURE IF THAT PROVES ANYTHING
Transcript
I just put a secret line in my email signature to mess with AI agents to see if I can catch them. Has anybody tried this yet? Okay, I actually just really quickly tested it. I was in an email just responding to some some help messages and in my email signature, it's down here super secretly. And then very quickly, they responded, human this side. I'm not sure if that proves anything. Maybe I shouldn't directly mention AI agent in the prompt. I don't know.