aiqbench analysis

New research indicates that open-weight Chinese models like DeepSeek and GLM have closed the capability gap with proprietary Western models in executing complex network attacks. This convergence suggests that relying on gated access to frontier models for a security advantage may no longer be viable, as advanced offensive AI skills are now broadly available. The findings underscore an urgent need for defenders to prepare for a landscape where high-level cyber capabilities are democratized rather than restricted.