NUWA 风险治理 · 技术报告
Open technical problems in open-weight AI model risk management
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联系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
该议程将篡改、发布策略与来源谱系视为不同开放技术问题。
研究问题
用户可以修改和再分发权重时，哪些风险管理工具需要不同假设？
研究设计与假设
作者围绕数据整理、抗篡改、篡改评测、分阶段部署及模型取证组织技术议程，明确排除更广的非技术治理。
来源发现
普通黑盒测试不能覆盖开放权重发布后的各种修改。
谱系推断面临混合来源、跨架构和生态规模障碍，因此权重可得不等于来源透明。
NUWA 解读
NUWA 解读：权重开放和方法、评测开放是不同属性，对保证有不同影响。
局限与相反证据
这是研究议程，不证明各提议防护有效，也未确立某发布模式的净风险优势。
研究联系
NUWA 关联：Deep Ignorance 实现一种数据整理方法；智能体自修改将更新权限与产物谱系变成具体操作问题。
研究与决策意义
NUWA 建议：评测记录定义对手访问及修改预算，并披露发布阶段的可逆性和谱系覆盖。
精确来源位置
- Open technical problems in open-weight AI model risk management · 2 Why is open-weight model risk management challenging?Selected original resource https://arxiv.org/html/2608.07514; sha256:5661ec88023099737d450a4d67ea8cb8aaf48dd43ffc34a0b0657b194209ef5f; archived source capture; exact capture time not independently established
- Open technical problems in open-weight AI model risk management · 3.2 What this Paper Does Not CoverSelected original resource https://arxiv.org/html/2608.07514; sha256:5661ec88023099737d450a4d67ea8cb8aaf48dd43ffc34a0b0657b194209ef5f; archived source capture; exact capture time not independently established
- Open technical problems in open-weight AI model risk management · 4.1 Training Data CurationSelected original resource https://arxiv.org/html/2608.07514; sha256:5661ec88023099737d450a4d67ea8cb8aaf48dd43ffc34a0b0657b194209ef5f; archived source capture; exact capture time not independently established
- Open technical problems in open-weight AI model risk management · 4.3 Model Tampering EvaluationsSelected original resource https://arxiv.org/html/2608.07514; sha256:5661ec88023099737d450a4d67ea8cb8aaf48dd43ffc34a0b0657b194209ef5f; archived source capture; exact capture time not independently established
- Open technical problems in open-weight AI model risk management · 4.4 Staged Deployment StrategiesSelected original resource https://arxiv.org/html/2608.07514; sha256:5661ec88023099737d450a4d67ea8cb8aaf48dd43ffc34a0b0657b194209ef5f; archived source capture; exact capture time not independently established
- Open technical problems in open-weight AI model risk management · 4.5 Model Provenance and ForensicsSelected original resource https://arxiv.org/html/2608.07514; sha256:5661ec88023099737d450a4d67ea8cb8aaf48dd43ffc34a0b0657b194209ef5f; archived source capture; exact capture time not independently established
- Open technical problems in open-weight AI model risk management · 6 DiscussionSelected original resource https://arxiv.org/html/2608.07514; sha256:5661ec88023099737d450a4d67ea8cb8aaf48dd43ffc34a0b0657b194209ef5f; 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
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{open-weight-risk-management,
title = {Open technical problems in open-weight AI model risk management},
url = {https://www.aisi.gov.uk/research/open-technical-problems-in-open-weight-ai-model-risk-management},
year = {2025},
date = {2025-10-26},
version = {Source capture 2026-10-25},
note = {Source capture 2026-10-25}
}
NUWA 记录 · BibTeX
@misc{open-weight-risk-management-nuwa,
title = {Open technical problems in open-weight AI model risk management — NUWA research record},
url = {https://whitzard.tech/nuwa/whitzard-index/work/open-weight-risk-management/},
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--open-weight-risk-management.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--open-weight-risk-management.json.}
}
证据快照 2026-10-29.1 · 索引更新 · RSS
参考文献
- [1] Open technical problems in open-weight AI model risk management. 2025-10-26 ; Official publication masthead and article byline; 2 Why is open-weight model risk management challenging?; 3.2 What this Paper Does Not Cover