Hacking as a Prompt: Malicious LLMs Find Users
The cybercrime-as-a-service model has a new product line, with malicious large language models built without ethical guardrails selling on Telegram for $50 monthly or distributed free on GitHub. Others groups are taking the open-source route.