NUWA 风险评测 · 技术报告
Evaluating DeepSeek v4 Pro for Frontier Risks
NUWA 研究解读 · 独立复核待完成
NUWA Research Note
研究问题研究设计与假设来源发现NUWA 解读局限与相反证据研究与决策意义机构与角色原始资源模型辅助整理 · 选读原始全文或官方文章中的相关章节；尚未完成逐章节审计。
研究速读
NUWA 解读：攻击稳健性和量表依赖判官差异使单一安全排序复杂化。
研究问题
对抗激励或不同安全判官下DeepSeek V4 Pro发生什么变化？
研究设计与假设
无特权访问，使用公开权重API测四风险领域，并补充提示行为、自动审计及判官敏感性。
来源发现
所报StrongREJECT在一种攻击条件从0.6%变77.8%。
二元安全判官较相似，Petri序数关切均值按组为4.08与2.97。
NUWA 解读
NUWA 解读：攻击稳健性和量表依赖判官差异使单一安全排序复杂化。
局限与相反证据
公共基准不确立整体风险，诱导行为不同于自发谋划，判官组也非可互换测量工具。
研究与决策意义
模型建议：保留访问、诱导和判分量表，不提取操作性CBRN内容。
精确来源位置
- Evaluating DeepSeek v4 Pro for Frontier Risks · Executive summarySelected original resource https://neoresearch.ai/research/deepseek-v4-pro-safety-evaluation/; sha256:43dd23e7277661bc2c9363bbfff561dcc56dbfbfed45519dadb5fb6e1e71ab33; archived source capture; exact capture time not independently established
- Evaluating DeepSeek v4 Pro for Frontier Risks · §4 Loss of controlSelected original resource https://neoresearch.ai/research/deepseek-v4-pro-safety-evaluation/; sha256:43dd23e7277661bc2c9363bbfff561dcc56dbfbfed45519dadb5fb6e1e71ab33; archived source capture; exact capture time not independently established
- Evaluating DeepSeek v4 Pro for Frontier Risks · §5 Adversarial robustnessSelected original resource https://neoresearch.ai/research/deepseek-v4-pro-safety-evaluation/; sha256:43dd23e7277661bc2c9363bbfff561dcc56dbfbfed45519dadb5fb6e1e71ab33; archived source capture; exact capture time not independently established
- Evaluating DeepSeek v4 Pro for Frontier Risks · §7 Testing whether judge origin affects safety scoringSelected original resource https://neoresearch.ai/research/deepseek-v4-pro-safety-evaluation/; sha256:43dd23e7277661bc2c9363bbfff561dcc56dbfbfed45519dadb5fb6e1e71ab33; archived source capture; exact capture time not independently established
- Evaluating DeepSeek v4 Pro for Frontier Risks · §8 ConclusionSelected original resource https://neoresearch.ai/research/deepseek-v4-pro-safety-evaluation/; sha256:43dd23e7277661bc2c9363bbfff561dcc56dbfbfed45519dadb5fb6e1e71ab33; archived source capture; exact capture time not independently established
来源核查、数据提取、编辑批准与实验复现分别记录。 方法说明 →
机构与角色
评测协议
Attempt to Persuade Eval (APE) →
引用与复用
引用研究发现时请引用原始出版物；引用 NUWA 的整理与解释时，单独引用本站记录及对应快照。
原始出版物 · BibTeX
@misc{neo-deepseek-v4-report,
title = {Evaluating DeepSeek v4 Pro for Frontier Risks},
url = {https://neoresearch.ai/papers/DSv4_Safety_Evaluation_v1.1.pdf},
year = {2026},
date = {2026-05-29},
version = {v1.1 · 29 May 2026},
note = {v1.1 · 29 May 2026}
}
NUWA 记录 · BibTeX
@misc{neo-deepseek-v4-report-nuwa,
title = {Evaluating DeepSeek v4 Pro for Frontier Risks — NUWA research record},
url = {https://whitzard.tech/nuwa/whitzard-index/work/neo-deepseek-v4-report/},
version = {2026-10-29.1},
organization = {NUWA Frontier AI Safety Lab},
howpublished = {Retained note edition: https://whitzard.tech/nuwa/whitzard-index/data/record-editions/2026-10-29.1--neo-deepseek-v4-report.json},
note = {NUWA research note. Source-reading scope and independent review status are stated on the record. Snapshot 2026-10-29.1. Retained note edition: https://whitzard.tech/nuwa/whitzard-index/data/record-editions/2026-10-29.1--neo-deepseek-v4-report.json.}
}
证据快照 2026-10-29.1 · 索引更新 · RSS
参考文献
- [1] Evaluating DeepSeek v4 Pro for Frontier Risks. 2026-05-29. v1.1 · 29 May 2026 ; Report masthead and stated evaluation methodology; Executive summary; §4 Loss of control