The underground market for malicious large language models (LLMs) is thriving, according to researchers from Indiana University Bloomington. They found 212 malicious LLMs for sale from April through September 2024.