Chinese LLMs Broaden the Gap Between Attackers & Defenders

Chinese open-weight AI models like GLM 5.2 and 360 Security Technology's Tulongfeng are demonstrating strong capability and cost-efficiency at finding software vulnerabilities, outperforming some frontier models on benchmarks and reportedly discovering thousands of vulnerabilities; this raises concerns that readily available, locally runnable models will accelerate attackers' ability to find and exploit weaknesses unless defenders rapidly prioritize patching, improve visibility and harden configurations.