A curated list of open-weight uncensored AI models is available for authorized red team operations, penetration testing, and security research. The collection, updated as of September 2026, includes models that are either uncensored or specifically fine-tuned for cybersecurity tasks.
Curated list of open-weight uncensored AI models for security research
The models, hosted on Hugging Face, range from small language models (SLM) to larger architectures including Qwen and DeepSeek variants. The list includes specialized versions, such as "abliterated" models, designed to bypass certain safety constraints to assist in offensive security testing and research. Use of these models is intended for authorized security research and education only.
Sources
- Uncensored and Offensive Security AI Models Benchmark （Hacker News Frontpage, 2026-09-29）