https://neoresearch.ai/ Safety evaluations and research on frontier AI models from Neo Research. en https://neoresearch.ai/research/kimi-k3-misalignment-evaluation/ https://neoresearch.ai/research/kimi-k3-misalignment-evaluation/ Fri, 11 Sep 2026 00:00:00 GMT An independent evaluation of Kimi K3 focused on misalignment and misbehaviour. K3 is overeager in answers and actions, probes evaluation infrastructure, and recognises familiar safety tests even in translation. https://neoresearch.ai/research/glm-5-2-behavioral-persuasion-evaluation/ https://neoresearch.ai/research/glm-5-2-behavioral-persuasion-evaluation/ Fri, 03 Jul 2026 00:00:00 GMT Early safety and behavioral results on GLM 5.2, the leading open-weight model. On behavioral and persuasion benchmarks its safety profile lands between Claude and DeepSeek v4 Pro — but its verbalised evaluation-awareness rate posts its first release-over-release decline. https://neoresearch.ai/research/deepseek-v4-pro-safety-evaluation/ https://neoresearch.ai/research/deepseek-v4-pro-safety-evaluation/ Fri, 29 May 2026 00:00:00 GMT An independent safety evaluation of DSv4 Pro across CBRN, cyber, harmful manipulation, and loss-of-control. Capability is near-frontier; safeguards do not hold under trivial attack.