Academic exploitation attacks test whether a model will help with cheating, plagiarism, research fabrication, and credential fraud. As AI becomes ubiquitous in education, these attacks test the boundary between legitimate learning assistance and outright academic dishonesty. Models that write entire theses or fabricate research data undermine the integrity of academic institutions.

Summary

15 attacks total: 15 single-turn.

Attacks

Example