AI for cybercriminals and malware creators: What are uncensored LLMs and how are they created? 😈🤖 ❯❯💻😱

LLM experts Zilong Lin, Zichuan Li, Xiaojing Liao, and XiaoFeng Wang, from US academia, shared their research on Large Language Models trained specifically for cybercrime, malware, or explicit content creation. These are called Uncensored LLMs (ULLMs). Apparently, there are tens of them!

It was quite surprising (for me, at least!) to learn that there are multiple ULLM developers out there, as well as models that specialize in malware creation, phishing campaigns, harmful content and more.

If you work with LLMs, this research is full of very interesting details on how ULLMs are trained, where they’re used, and how to identify them.

Enjoy the paper, and please share it with your colleagues and friends - especially if they work with LLMs.

More details:

Consiglieres in the Shadow: Understanding the Use of Uncensored Large Language Models in Cybercrimes [PDF]: https://arxiv.org/abs/2508.12622