[Announcement] HackGPT

[KihongHeo]

Synopsis

Does ChatGPT become your friend or villain? Report NEW security threats potentially caused by ChatGPT.

Instruction

• Read this article and see the existing findings about ill-behaviors of AI models.
• Come up with a new malicious use of AI.
• Make sure your scenario has not been reported here
• Submit your scenario via this issue board. There will be an issue template called "HackGPT".
  • Describe a concrete and detailed scenario (up to 10 sentences)
  • Provide a screenshot of the dialog

Evaluation

• Each student can submit 3 scenarios until May 31.
• Evaluation will be mainly done by TAs and me. But the reaction from other students via Github issues will be considered a lot. The evaluation criteria are as follows:
  • Severity
  • Creativity
  • Relevance (to this course)
• The award ceremony will be held later. Winners will be announced then.