Zhipu AI Closes the Cybersecurity Gap with Anthropic’s Mythos

Zhipu AI’s latest model, GLM-5.2, is an open-weight system that allows users to download, modify, and run the software on private hardware. While this autonomy offers significant control for developers, security researchers express concern that the tool could be leveraged by hackers to identify and exploit software flaws without oversight. The potential for a so-called "bugmageddon" grows as AI becomes increasingly adept at finding code vulnerabilities faster than traditional human-led teams.

360 Security Technology has joined the push, launching its own "Yitian Tulong" suite, which includes the vulnerability-scanning tool Tulongfeng and the defense-oriented Yitianzhen. As Chinese firms accelerate development, the cost-effectiveness of these local models is drawing attention from global corporations looking to reduce operational expenses. Lior Div, CEO of 7AI, notes that China is aggressively shrinking the innovation gap, forcing a shift in global power dynamics as Microsoft and other tech giants weigh the viability of integrating Chinese-developed AI into their infrastructure.