Backdoor Unlearning Generalization: A Path toward the Removal of Unknown Triggers in LLMs

Backdoor attacks in Large Language Models LLMs are a growing security concern, where models can generate adversary-chosen content. Existing defenses target backdoors one at a time and typically require knowledge of the trigger, leaving the defender at a structural disadvantage when unknown...

📄 Espanso 2.3.0 Configuration Injection

This Python script is a configuration manipulation tool for Espanso version 2.3.0 that modifies its YAML configuration file base.yml to add new text triggers capable of executing system commands via shell or script extensions...

PT-2026-44062

Name of the Vulnerable Software and Affected Versions Budibase versions prior to 3.39.0 Description An issue exists in the open-source low-code platform where the webhook schema-building endpoint is registered under builderRoutes, but the generic authorization middleware fails to enforce...

Budibase 安全漏洞

Budibase is an open-source platform developed by Budibase in the UK. It allows for the creation of internal applications, workflows, and management panels within minutes. Versions of Budibase prior to 3.38.1 contained security vulnerabilities. These vulnerabilities stemmed from action triggers th...

Malicious code in fca-official-uzair-rajput (npm)

--- -= Per source details. Do not edit below this line.=- Source: amazon-inspector 83c96ed99bb1a48e80228ec0ca012c1dbb7817fe1dbbd492fcb3d2927805f29e fca-official-uzair-rajput is a Facebook chat API library whose only documented entry point, login, invokes an auto-update routine on every call when...

Incorrect Authorization

Overview github.com/mattermost/mattermost/server/channels/app is a private-cloud Slack alternative Affected versions of this package are vulnerable to Incorrect Authorization via the command update API. An attacker can impersonate existing system or custom commands by editing their own slash...

Incorrect Authorization

Overview Affected versions of this package are vulnerable to Incorrect Authorization via the command update API. An attacker can impersonate existing system or custom commands by editing their own slash command trigger to match an already-registered trigger, potentially hijacking command...

PT-2026-41649

Mattermost versions 11.5.x = 11.5.1, 10.11.x = 10.11.13, 11.4.x = 11.4.3 Fail to enforce slash command trigger-word uniqueness during command updates which allows an authenticated team member with Manage Own Slash Commands permission to hijack and impersonate existing system or custom slash...

Mattermost 安全漏洞

Mattermost is an open-source collaboration platform developed by the American company Mattermost. Versions of Mattermost such as 11.5.1 and earlier 11.5.x series, 10.11.13 and earlier 10.11.x series, and 11.4.3 and earlier 11.4.x series have security vulnerabilities. These vulnerabilities stem fr...

Can a Single Message Paralyze the AI Infrastructure? the Rise of AbO-DDoS Attacks through Targeted Mobius Injection

Large Language Model LLM agents have emerged as key intermediaries, orchestrating complex interactions between human users and a wide range of digital services and LLM infrastructures. While prior research has extensively examined the security of LLMs and agents in isolation, the systemic risk of...

CVE-2026-44694

n8n-MCP is an MCP server that provides AI assistants access to n8n node documentation, properties, and operations. From version 2.18.7 to before version 2.50.2, there is an authenticated server-side request forgery vulnerability affecting the webhook trigger tools, the n8n API client N8NAPIURL, a...

Astra Linux - уязвимость в linux-5.10

In the Linux kernel, the following vulnerability has been resolved: Tracing: The “Drain deferred trigger” operation is freed if kthread creation fails. Registration of boot-time triggers may fail before the trigger-data cleanup is completed. If a kthread exists, deferring those operations until...

Astra Linux - уязвимость в linux-6.1

In the Linux kernel, the following vulnerabilities have been resolved: Tracing: Fixed an issue where corrupted data in the “hist” field could corrupt the “namedTriggers” list. The following commands cause a crash: cd /sys/kernel/tracing/events/rcu/rcucallback echo...

Important: kernel6.12

Issue Overview: In the Linux kernel, the following vulnerability has been resolved: netfilter: ctnetlink: remove refcounting in expectation dumpers CVE-2025-39764 In the Linux kernel, the following vulnerability has been resolved: ipv6: use RCU in ip6xmit CVE-2025-40135 In the Linux kernel, the...

DETOUR: A Practical Backdoor Attack against Object Detection

Object detection OD is critical to real-world vision systems, yet existing backdoor attacks on detection transformers DETRs for OD tasks rely on patch-wise triggers optimized at fixed locations with minimal perturbations. Such attacks overlook that backdoor triggers in the real world may appear a...

CVE-2026-31481

In the Linux kernel, the following vulnerability has been resolved: tracing: Drain deferred trigger frees if kthread creation fails Boot-time trigger registration can fail before the trigger-data cleanup kthread exists. Deferring those frees until late init is fine, but the post-boot fallback mus...

CVE-2026-33594

A client can trigger excessive memory allocation by generating a lot of queries that are routed to an overloaded DoH backend, causing queries to accumulate into a buffer that will not be released until the end of the connection...

False Security Confidence in Benign LLM Code Generation

Prior work has demonstrated that functionally correct yet vulnerable outputs arise systematically in threat-oriented settings, where adversarial or implicit channels are used to induce security failures in code agents and automated patching workflows. This note introduces a complementary but...

Unity Linux 20.1050e / 20.1060e / 20.1070e Security Update: kernel (UTSA-2026-007246)

The Unity Linux 20 host has a package installed that is affected by a vulnerability as referenced in the UTSA-2026-007246 advisory. In the Linux kernel, the following vulnerability has been resolved: tracing/histograms: Add histograms to histvars if they have referenced variables Hist triggers ca...

Your LLM Agent Can Leak Your Data: Data Exfiltration Via Backdoored Tool Use

Tool-use large language model LLM agents are increasingly deployed to support sensitive workflows, relying on tool calls for retrieval, external API access, and session memory management. While prior research has examined various threats, the risk of systematic data exfiltration by backdoored...