All posts
Topic · 4 articles
Model Risk: articles and guides
4 PlayCISO articles on model risk, newest first. The most recent is “Dolphin3-Cyber-8B: What It Is and the Security Risk (2026)”.
Dolphin3-Cyber-8B: What It Is and the Security Risk (2026)
Dolphin3-Cyber-8B is a community cybersecurity fine-tune built on the abliterated (refusal-removed) Dolphin3.0-Llama3.1-8B base. What it actually is, what its publishers claim, how it runs locally, and how to assess the risk if it shows up on your network.
October 27, 2026
Abliterated Models Explained: Meaning and How They Work
Abliteration removes an open-weights LLM’s ability to refuse by deleting its refusal direction. Not a jailbreak or a retrain. How it works and the risk.
October 12, 2026
Open-Weight Model Security Standard: A Checklist
Baseten, Hugging Face and Goodfire are building safety infrastructure for open-weight models. A checklist of what a credible security standard needs.
September 24, 2026
Dolphin3-Cyber-8B GGUF: What It Is and the Risks
Dolphin3-Cyber-8B is a free Llama 3.1-based 8B Hugging Face model fine-tuned from an abliterated base on security data; GGUF builds start at 3.18GB. Risks.
September 22, 2026