Malicious LLMs

Working with UVA Research to detect malicious, 'jailbroken' large language models on dark markets.