China's GLM-5.2 AI model bypasses US export controls, raises security fears

A new open-weight AI model, GLM-5.2, released by China-based Z.ai, poses a national security concern due to its powerful coding and vulnerability discovery capabilities. Unlike U.S. models like Anthropic's Mythos and OpenAI's GPT-5.6, which are subject to export controls and restricted access, GLM-5.2 is freely downloadable and can be run on private hardware, leaving no usage record. This lack of containment means the model can be used by malicious actors without oversight, mirroring the capabilities that led U.S. officials to restrict access to advanced American AI models. AI

IMPACT Accelerates the proliferation of powerful AI capabilities, challenging existing national security frameworks and export controls.

RANK_REASON A new AI model release from a non-US lab with significant implications for national security and international AI policy. [lever_c_demoted from significant: ic=1 ai=1.0]

Read on Forbes — Innovation →

• Andy Jassy
• Anthropic
• Claude Opus 4.8
• Dario Amodei
• GLM-5.2
• GPT-5.6
• Howard Lutnick
• MIT license
• Mythos
• OpenAI
• Scott Bessent
• Z.ai

• model release
• policy
• safety

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →