Developing story · 2 episodes · 28 Sept to 29 Sept
Zhipu GLM-5.3 Automation Security Crisis
The saga tracks Zhipu's rapid deployment of GLM-5.3's autonomous infrastructure capabilities, which were subsequently flagged by Anthropic for severe security vulnerabilities. The story currently stands at a critical juncture where the model's advanced self-improvement features are being scrutinized for their potential to hijack code execution.
-
Anthropic’s Frontier Red Team finds GLM-5.3 and Claude Mythos Preview can hijack code
Anthropic’s Frontier Red Team tested two models—GLM-5.3 and Claude Mythos Preview—on 100 tasks from its internal Binary Exploitation benchmark. GLM-5.3 succeeded in 4% of trials, while Claude Mythos…
1 source -
Zhipu automates infrastructure with GLM-5.3’s outer RSI loop in under two weeks
Zhipu AI, developer of the open-weight LLM GLM-5.3, detailed how it used its own model to accelerate internal infrastructure work. The team deployed an Infra Agent powered by GLM-5.3 to optimize…
3 sources primary source
Who and what
Zhipu AIGooglePlanetPhysical IntelligencePeriodic LabsAnthropicGLM-5.3GLM-5.3-FlashKimi 2.6GPT-6 AstraClaude Fable 5.1Periodic Neon