The tendency of code-generating large language models (LLMs) to produce completely fictitious package names in response to certain prompts is significantly more widespread than commonly recognized, a ...