AI refusal mechanisms
Safety protocols embedded in AI models to reject harmful or dangerous prompts, trained through red-teaming and fine-tuning but reliant on probabilistic activations that are not fully understood.
Catégorie : Tech
Articles liés (1)

