Prompt Override

Prompt Override is a hacking-themed serious game where the player is a novice hacker collaborating with a rogue Large Language Model (LLM) to infiltrate and eventually dismantle a criminal corporation. The game showcases (through a real LLM-based simulation) real-world risks of prompt injection, as the player's carefully crafted inputs trick the antagonist LLM to ignore its system prompt leading to leaked data or misinformation, and model misuse.

Relevant Publications

  • Roberto Gallotta, Antonios Liapis, and Georgios N. Yannakakis: "Prompt Override: LLM Hacking as Serious Game," in Proceedings of the IEEE Conference on Games, 2025. PDF BibTex