In his Note, Trent Kannegieter reframes questions of AI liability around how AI systems work in practice. He argues that understanding LLMs’ key technical features—especially in-production nondeterminism—enables tort-law redress when LLM-based products cause harm.
