The International AI Security Benchmark CyberGym released its latest ranking today. The Chinese team Dark Navy (DARKNAVY) jointly developed DoGNAVY, which...
行业动态IT之家(RSS)
Today AI Intelligence Brief
The International AI Security Benchmark CyberGym released its latest ranking today. The Chinese team
Dark Navy (DARKNAVY) jointly developed DoGNAVY, which ranked third globally and first among open-source projects with a pass rate of 90.8%. DoGNAVY only used a single open-source general model (GLM-5.2, open-sourced by Zhipu in June), while the higher-ranked Microsoft and Google 'operate' using a combination of multiple top-level closed-source models. CCTV News reported that AI security offense and defense is entering a practical stage, and the open-source route is enabling top security capabilities to be mastered by more innovators.
该排名可能提升业界对开源模型用于 AI 安全攻防的关注,并推动更多团队评估自主部署路线。但基准排名只反映特定测试条件,不能直接推导真实环境中的整体安全能力。 值得关注 CyberGym 后续是否公开更完整的测试任务、评分方法、复现条件与各方案配置,也应观察 DoGNAVY 在其他安全基准及真实攻防场景中的表现是否保持一致。
Source and Copyright
The readable text on this page was extracted from the public source and organized with attribution, publication time and the original link. Copyright remains with the original author and publisher.
The International AI Security Benchmark CyberGym released its latest ranking today. The Chinese team Dark Navy (DARKNAVY) jointly developed DoGNAVY, which...
IT之家(RSS)2026-08-05T08:19:31.000Z
Scan to open this article
Aioga aggregates global AI updates and preserves source information for verification and citation.