AI security & privacy, algorithmic bias, foundations of ML
Video-LLM evaluations on CinePile's evaluation split.