Large Language Models (LLMs) have a serious “package hallucination” problem that could lead to a wave of maliciously-coded packages in the supply chain, researchers have discovered in one of the ...