2026-09-02 归档

今日共收录 1494 条安全情报，包含 600 个 CVE，479 篇安全通告，以及 120 篇研究论文。

安全公告2026-09-01 发布INFO

Ash.Policy.Authorizer returns records denied by a runtime read policy to any actor

CVE-2026-82747

Incorrect Authorization vulnerability in ash-project ash returns records that a runtime read policy denies to any actor.

历史摘要 · 未重新核验 · OSV.dev

#advisory #cwe-863

安全公告2026-09-01 发布INFO

Ash.Filter.Runtime materializes a combinatorial cross-product over to-many relationships, exhausting memory

CVE-2026-82742

Uncontrolled Resource Consumption vulnerability in ash-project ash lets an attacker exhaust node memory by matching a fi

历史摘要 · 未重新核验 · OSV.dev

#advisory #cwe-400

安全公告2026-09-01 发布INFO

Authorization-redacted field values disclosed through AshTypescript result normalization

CVE-2026-82730

Incorrect Authorization vulnerability in ash-project ash_typescript allows an unauthorized RPC caller to read attribute

历史摘要 · 未重新核验 · OSV.dev

#advisory #cwe-863

安全公告2026-09-02 发布INFO

PLANET VDR-300NU ADSL Router - 未授权修改DNS

历史摘要 · 未重新核验 · 知道创宇 / Seebug

#knownsec #seebug #vulnerability

安全公告2026-09-02 发布INFO

Acknowledgements

历史摘要 · 未重新核验 · Android Security Bulletin

#android #mobile

漏洞预警2026-09-01 发布INFO

Next.js is a React framework for building full-stack web applications. From 13.4.0 until 15.5.24 and 16.3.3, Next.js applications using Pages Router or App Router without Cache ...

CVE-2026-75604

Next.js is a React framework for building full-stack web applications. From 13.4.0 until 15.5.24 and 16.3.3, Next.js app

历史摘要 · 未重新核验 · NVD

#cve #cwe-22

漏洞预警2026-09-01 发布INFO

A vulnerability was detected in Cobham SATCOM VSAT7090 Maritime Satellite Router up to 20260704. This issue affects the function c_set_reports_decode of the file mail-report.sh ...

CVE-2026-83772

A vulnerability was detected in Cobham SATCOM VSAT7090 Maritime Satellite Router up to 20260704. This issue affects the

历史摘要 · 未重新核验 · NVD

#cve #cwe-74 #cwe-77

学术论文2026-09-02 发布INFO

Alleviating the Fear of Losing Alignment in LLM Fine-tuning.

本文研究大型语言模型（LLM）在微调过程中对齐（alignment）能力意外丧失的问题。尽管对齐训练可以促使模型拒绝回答不道德或有害的问题，但当模型针对下游任务进行微调时，这种对齐可能被削弱，导致模型再次输出有害内容。作者观察到，一个经过对

历史摘要 · 未重新核验 · IEEE S&P

#llm #fine-tuning #alignment #model-security

学术论文2026-09-02 发布INFO

Comet: Accelerating Private Inference for Large Language Model by Predicting Activation Sparsity.

本文提出一种名为 Comet 的高效隐私推理系统，旨在加速基于云平台的大语言模型（LLM）推理服务，同时缓解用户对敏感信息泄露的担忧。当前主流方案采用安全多方计算（MPC）保护推理过程中的数据隐私，但 MPC 需要频繁的服务器间通信，导致显

历史摘要 · 未重新核验 · IEEE S&P

#agent-security #llm #mpc #privacy-inference

安全公告2026-09-02 发布INFO

A vulnerability was identified in MapQuest Get Directions App 10.16.1 on Android. This...

CVE-2026-84442

A vulnerability was identified in MapQuest Get Directions App 10.16.1 on Android. This vulnerability affects the functio