China’s Zhipu AI GLM-5.2 Matches Claude Mythos in Vulnerability Detection Tasks

Zhipu AI’s open-weight GLM-5.2 model is performing on par with Anthropic’s restricted Claude Mythos on specific cybersecurity and software vulnerability detection tasks, a development that is intensifying concerns within the U.S. government about the effectiveness of its AI export control strategy. Released on June 13, 2026, under a permissive open-weight license, GLM-5.2 allows any researcher […]

The post .