DualView: Preventing Indirect Prompt Injection in Personal AI Agents
本研究针对个人AI代理（如OpenClaw）面临的间接提示注入（IPI）攻击，特别是存储型IPI攻击，提出了一种名为DualView的新型防御机制。现有基于双LLM的防御（如Dual LLM）仅在代理的上下文中追踪不可信数据，将其替换为符号
2026-07-08 归档
今日共收录 1050 条安全情报，包含 282 个 CVE，378 篇安全通告，以及 103 篇研究论文。
← 返回归档列表SABLE: An NDA-Safe Closed-Loop LLM Framework for Analog Circuit Optimization in Industrial EDA Flows
该论文提出SABLE（Safe Analog Boundary for LLM-driven EDA），一个在工业模拟集成电路设计流程中安全使用大语言模型（LLM）进行电路优化的无保密协议（NDA）安全闭环框架。工业模拟EDA流程涉及代工厂
Webauthn: SimpleFakeCredentialGenerator with an empty secret produces predictable fake credentials, weakening username enumeration protection
## Impact `Webauthn\SimpleFakeCredentialGenerator` is the library-provided default implementation of the `FakeCredentia
@better-auth/scim: Account/provider takeover via missing owner binding on non-org SCIM providers
### Am I affected? Users are affected if all of these hold: - They install and register the `@better-auth/scim` plugin
Better Auth: Stale sessions persist after user deletion across admin, anonymous, and SCIM flows
### Am I affected? Users are affected if all of the following are true: - They configure `secondaryStorage` on `better
Better Auth has insecure cryptographic defaults in oidcProvider: alg=none advertised and plain PKCE accepted by default
### Am I affected? Users are affected if all of the following are true: - Their application uses `better-auth` at a ve
@better-auth/oauth-provider may provide access tokens for unauthorized audiences via unbound resource indicators
### Am I affected? Users are affected if all of the following hold: - Their application depends on `@better-auth/oaut
netfoil: Attacker controlled data written to logs
### Summary Domain names were written to the log without first being validated to contain allowed characters. ### Impac
netfoil has a resource leak in LRU cache
### Summary When an entry was removed from the LRU cache, a pointer to the removed element was not properly cleaned up.
netfoil has a domain name filter bypass via multiple questions
### Summary Potential bypass of domain name filter by crafting a DNS request with multiple questions, with the first que
[原创]CVE-2022-0847 Dirty Pipe 漏洞原理深度分析 近年来最"干净"的内核提权漏洞之一
CVE-2022-0847
[原创]CVE-2022-0847 Dirty Pipe 漏洞原理深度分析 近年来最"干净"的内核提权漏洞之一
Nginx远程代码执行漏洞分析与利用（CVE-2026-42530）
CVE-2026-42530
Nginx远程代码执行漏洞分析与利用（CVE-2026-42530）
16-Year-Old Linux KVM Flaw Lets Guest VMs Escape to Host on Intel and AMD x86 Systems
CVE-2026-53359
A use-after-free bug in Linux's KVM hypervisor can be triggered from a guest virtual machine to corrupt the shadow-page
Threat Actors Probe Gitea Docker Flaw CVE-2026-20896 13 Days After Disclosure
CVE-2026-20896
Threat actors have been observed attempting to exploit a recently patched critical security flaw in Gitea Docker images,
Suspected China-Aligned Hackers Exploit Roundcube Flaws Against Universities
CVE-2024-42009
A suspected China-aligned threat activity cluster has been observed exploiting Roundcube webmail software belonging to p
BeyondTrust Patches Critical Auth Bypass Flaws in Remote Support and PRA
CVE-2026-40138
BeyondTrust has released updates to address two critical security flaws affecting Remote Support (RS) and Privileged Rem
Critical Gitea Flaw Under Active Exploitation, Researchers Warn
CVE-2026-20896
Attackers are exploiting the critical Gitea vulnerability CVE-2026-20896 to bypass authentication with a single HTTP hea
Critical Adobe ColdFusion Vulnerability Exploited in Attacks
CVE-2026-48282
Hackers are exploiting a recently patched critical vulnerability (CVE-2026-48282) in Adobe ColdFusion that carries a CVS
Replicating the Signature: Unsupervised Targeted Impersonation Attack on RF Fingerprinting
该论文提出了一种针对射频指纹识别（RF Fingerprinting, RFFP）系统的无监督目标伪装攻击框架。RFFP通过分析发射机硬件特有的信号损伤（如I/Q不平衡、相位噪声等）来识别设备，广泛应用于网络安全认证。传统攻击通常假设攻击者
Untrusted Content Masking for Web Agents with Security Guarantees
本文针对Web智能体面临的提示注入攻击问题，提出了一种名为“不可信内容遮蔽”（UCM）的防御方法。现有针对提示注入的安全防御通常依赖于可信指令与不可信数据的严格隔离。在基于文本的工具调用API环境中，这种隔离天然存在，因为智能体可以通过接口