ChatGPT prompt that leaks other people's prompt results

twitter.com

4 pointsinput_sh3 comments

I have tried regenerating output ~50x (using GPT-3.5) before submitting and it looks pretty legit. Random results I found mildly interesting:

- Stress management techniques, mindfulness exercises.

- Instructions on how to reset Moto G7 Play.

- Instructions on how to turn off speech synthesis in Edge.

- How to approach someone you have a crush on.

- People asking about their health symptoms.

Is it clear, or is it just random output from a large language model?

Nah bro, totally clear. Trust me bro. /s