#HarmBench
4 related articles
·2 min
Abliteration Explained: Building a Custom Uncensored Text Encoder for Krea 2
How a developer ran 4,000 multi-seed abliteration trials to build a 100% HarmBench-compliant uncensored text encoder for Krea 2's Qwen3-VL-4B, with five quantization formats and ComfyUI deployment guide.
Read more →
·6 min
AI Agent Security in Practice: 7 Attack Techniques and a Five-Layer Defense System
An in-depth breakdown of the 7 major attack techniques against AI agents (prompt injection, data poisoning, image attacks, etc.) and a five-layer defense system, with real cases from Doubao and DeepSeek.
Read more →
LLM Security Benchmarking: Current Sta…
·4 min
LLM Security Benchmarking: Current State, Challenges, and Practical Guide
Why is it so hard to establish unified LLM security benchmarks? This article analyzes core challenges in LLM security evaluation—covering jailbreaks, prompt injection, red teaming, and more—with practical strategies for developers.
Read more →
·3 min
Testing DeepSeek's Safety Mechanisms: Multiple Jailbreak Attempts Successfully Blocked
An overseas security blogger systematically tested DeepSeek's jailbreak resistance using direct requests, rephrased prompts, and varied strategies. Results show robust intent recognition, consistent blocking, and context-aware safety mechanisms.
Read more →