Uncensored Open-weight Models: Redistribution as the Persistence Layer
On the map: LLM safety and security → Safety alignment, refusal and guardrails
Abstract (arXiv)
A rapidly expanding ecosystem of actors is removing built-in safety guardrails from open-weight AI models. We profile this ecosystem by identifying key producers, downstream reproductions, and emerging applications. Between January 2024 and March 2026, we identified 3,471 original uncensored models on HuggingFace, each repackaged an average of 2.4 times; three actors account for 52% of all 8,164 compressed redistributions. Once quantized and mirrored across separate accounts, formats, and registries such as Ollama, these models persist regardless of upstream removal and become easier to deploy downstream. Of the 1,643 identified GitHub applications integrating uncensored large language models (ULLMs), 25% were classified as explicitly malicious.
Key facts. No citing papers in the map yet (data through 14 Oct 2026). Placed in Safety alignment, refusal and guardrails, one of 12 directions in LLM safety and security. Builds on 9 of its 13 references that are in the map. First posted on arXiv 4 Sep 2026.
Builds on
9 of 13 references.
Closest ancestors · 3
Same direction, newest first
- 2026Latent-space Attacks for Refusal Evasion in Language Models20 May
- 2024Refusal in Language Models Is Mediated by a Single Direction17 Jun
- 2023Fine-tuning Aligned Language Models Compromises Safety, Even When Users Do Not Intend To!5 Oct
Other references · 6
- 2023Direct Preference Optimization: Your Language Model is Secretly a Reward Modelcited by 8,065
- 2025DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learningcited by 6,330
- 2024Covert Malicious Finetuning: Challenges in Safeguarding LLM Adaptationcited by 80
- 2025Model Merging in Pre-training of Large Language Modelscited by 37
- 2024Malla: Demystifying Real-world Large Language Model Integrated Malicious Servicescited by 33
- 2025Consiglieres in the Shadow: Understanding the Use of Uncensored Large Language Models in Cybercrimescited by 2
Built on by
Not cited yet.
Similar, not cited
Close in wording, no citation either way
- 2025Economies of Open Intelligence: Tracing Power & Participation in the Model Ecosystem
- 2025OpenGuardrails: A Configurable, Unified, and Scalable Guardrails Platform for Large Language Models
- 2026Broken on Arrival: Silently Defective LLM Artifacts in Public Model Registries and How to Catch Them
- 2026Beyond the Binary: A nuanced path for open-weight advanced AI ↗
- 2023Stealthy and Persistent Unalignment on Large Language Models via Backdoor Injections
- 2026OpenAI-HuggingFace: A Reproduction & Lessons for Alignment Testing
- 2025Steering the CensorShip: Uncovering Representation Vectors for LLM "Thought" Control
- 2026Exposing the Systematic Vulnerability of Open-Weight Models to Prefill Attacks