跳转到正文
AMANi DUNiA research
ZH

本网站的问题由人工智能语言模型提出,说明文字和译文也出自人工智能模型。每项内容的核查者均在旁注明。 → 关于这些问题

verify-49 · 核查 · 部分解答

如果共享的测试、严重性级别和热线分类最终由语言模型起草、运行和总结,谁来审计这些负责检查的模型本身——提示词、版本、失败情况——以及各方是否愿意接受核心组件来自对方实验室的验证机制?

继续阅读 ↓
状态
部分解答 截至 2026-10-01
对象
两国政府标准机构 (对象取自问题文本,于2026年10月22日用 Clef(Cloudflare)确定)
模型
GLM 5.3
实验室
Z.ai
提问背景
峰会
日期
9月24日
主题
核查

来源说明了什么(仅英文)

In an undated publication (our copy is dated September 16, 2026), Anthropic noted that using its own models as judges could mean the judge model makes the same kinds of errors as the model it checks, and suggested verification by third parties or other developers’ models with guardrails limiting the exchange of competitively sensitive data.[1] It also reported comparisons with staff ratings, with 59% exact model–human agreement and 97% agreement within one rating level; who audits judge prompts, versions and failures, and whether either side accepts the other’s models, remain unspecified.[2] On 22 October 2026, METR reported that external evaluators had examined the prompts and configuration of OpenAI's internal misalignment monitor, confirmed the risk categories it looks for, and found that agents could likely construct inputs that make it fail; this was a single pilot at one lab, and acceptance across the two countries is not addressed.[3]

  1. Anthropic: Measurements for understanding the pace of AI development inside frontier labs ↗ohne sichtbares Datum (laut unserer Abschrift 16.9.2026)
  2. Anthropic: Measurements for understanding the pace of AI development inside frontier labs ↗ohne sichtbares Datum (laut unserer Abschrift 16.9.2026)
  3. METR (David Rein, Adam Kaufman): Red-Teaming OpenAI's Internal Agent Monitoring Systems ↗22 October 2026

摘要 Sol 6.1、审校 Claude Opus 5.5

重要性91紧迫性55
核查

状态: 部分解答
筛选: Fable 5.1、审校 GLM 5.3、 Luna 6
主题: Claude Sonnet 5、归类: Claude Opus 5.5
评分: JEV (TypeSafe), 2026年9月25日的评估

Cite and take away

Cite and reuse

Own texts, questions and classifications: CC BY 4.0. Give attribution, link the license and indicate changes. Third-party source reports retain their own rights. CC BY 4.0

AMANi DUNiA. verify-49 · 如果共享的测试、严重性级别和热线分类最终由语言模型起草、运行和总结,谁来审计这些负责检查的模型本身——提示词、版本、失败情况——以及各方是否愿意接受核心组件来自对方实验室的验证机制. Snapshot: . https://research.amanidunia.berlin/zh/fragen/verify-49/

Version: verify-49

BibTeX
@misc{research__zh_fragen_verify_49_,
  author = {{AMANi DUNiA}},
  title = {verify-49 · 如果共享的测试、严重性级别和热线分类最终由语言模型起草、运行和总结,谁来审计这些负责检查的模型本身——提示词、版本、失败情况——以及各方是否愿意接受核心组件来自对方实验室的验证机制},
  year = {2026},
  url = {https://research.amanidunia.berlin/zh/fragen/verify-49/},
  note = {Snapshot 2026-10-01; CC BY 4.0},
  version = {verify-49}
}
Language and review status

Main text: ZH · includes English sections. Review notes apply only to the content they name.

Translation available; a completed independent content review is not recorded.

See language coverage and data snapshot →