Loading paper
AlignPoison: Conditional Targeted Poisoning of Generative Recommenders via Decoding-Aligned Prefixes — Minlit