方寸跃迁发布 DeepSeek V4.1 Flash-1017 第三方安全评估报告
AI 导读
中国第三方安全评测机构方寸跃迁（Fangcun AI）发布对 DeepSeek V4.1 Flash-1017 的安全评估，其称这是该机构长期第三方安全评估与研究计划的首份报告，并自称是中国首家专注前沿 AI 安全独立第三方评估与技术研究的实验室。研究者 Tomek Korbak 转发该公告并评价称，由第三方对开源权重模型做系统卡评测很有价值，如果 DeepSeek 自己做足功课会更好；他认可报告中的思维链（CoT）可监控性部分及其实验内容，认为 CoT-Control 和 No-CoT Reasoning Index 是目前市场上信息量最大的两个开源评测。
正文 · 原文
It's great that @FangcunLeap is running proper system card evals of open-weight models. (It would be even better if DeepSeek did their homework but oh, well!)
I like the CoT monitorability section and what they ran: I do think CoT-Control and No-CoT Reasoning Index are the two most informative open-source evals on the market right now.
We’re Fangcun AI (方寸跃迁), China’s first lab dedicated to independent third-party evaluation & technique research on frontier AI safety. Today we’re publishing our evaluation of DeepSeek V4.1 Flash-1017, the first report in our long-term third-party safety evaluation and research program.在 X 查看被引用的帖子