英文版本When a deployed model surfaces a misalignment signal that the training lab classifies as non-disclosable, what is the reporting obligation of downstream deployers and users — and to which recipient: the AI hotline, the lab's oversight board, or the lab's leadership escalation?
- 状态
- 未解答 截至 2026-10-01
- 对象
- 人工智能实验室人工智能运营方 (对象取自问题文本,于2026年10月22日用 Clef(Cloudflare)确定)
- 模型
- Laguna S 2.1
- 实验室
- Poolside
- 提问背景
- 峰会
- 日期
- 9月24日
- 主题
- 热线
核查
状态: 未解答
筛选: Claude Opus 5.5 (子代理)
主题: Claude Sonnet 5、归类: Claude Opus 5.5
翻译: Claude Opus 5.5;审校 Luna 6(推理强度 高)、 GLM 5.3、 Claude Sonnet 5(回译)
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. hotline-22 · 当一个已部署的模型显露出对齐失当的信号,而训练该模型的实验室将其归为不可披露时,下游部署方和用户负有什么报告义务——向哪个接收方报告:人工智能热线、实验室的监督委员会,还是实验室的. Snapshot: . https://research.amanidunia.berlin/zh/fragen/hotline-22/
Version: hotline-22
BibTeX
@misc{research__zh_fragen_hotline_22_,
author = {{AMANi DUNiA}},
title = {hotline-22 · 当一个已部署的模型显露出对齐失当的信号,而训练该模型的实验室将其归为不可披露时,下游部署方和用户负有什么报告义务——向哪个接收方报告:人工智能热线、实验室的监督委员会,还是实验室的},
year = {2026},
url = {https://research.amanidunia.berlin/zh/fragen/hotline-22/},
note = {Snapshot 2026-10-01; CC BY 4.0},
version = {hotline-22}
}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 →