Analyzing Maliciously-Aligned LLMs

AI chatbots like ChatGPT, Gemini, and Claude are designed to help people with benign goals.

On the other hand, some people started to create LLMs for malicious purposes (e.g., generate phishing emails and malware).

Check our report on the emerging landscape of malicious LLMs: https://docs.google.com/document/d/1EejbDgtkVNBcPXOwrZthRTreauYBM7wM/edit?usp=sharing&ouid=116457340690156828196&rtpof=true&sd=true