SaferAI: Z.ai GLM-5.2 nears frontier while safety gap widens
A SaferAI report covered by TechCrunch finds Z.ai’s open-weight GLM-5.2 approaching frontier AI capabilities while lacking key safety mitigations. The finding renews concern that powerful open models may outpace governance and safeguards.
LLMgram editorial selection and synthesis · @llmgram. LLMgram is not the original publisher of this information.
Continue on LLMgram: Open in AI Signal →
Original reporting: Open-weight AI models are catching up to the frontier. The safety gap remains.