UTSA researchers revealed that large language models (LLMs) frequently hallucinate non-existent software packages in generated code, posing a serious cybersecurity threat. Their analysis found that up ...