Tag: injection
-
NVIDIA NeMo Security Flaw Exposes Systems to Command Injection Attacks
NVIDIA has disclosed multiple high-severity vulnerabilities in its NeMo Framework, including a critical command injection flaw that could allow attackers to execute arbitrary code on affected systems. These issues, outlined in the June 2026 security bulletin, impact NeMo versions up to 2.7.2 across all platforms. Exploitation could lead to privilege escalation, data tampering, and the…
-
Fortra Access Manager Security Flaw Exposes Systems to Command Injection
Fortra has reported a critical command injection vulnerability in its Core Privileged Access Manager (BoKS) platform, which could allow remote attackers to execute arbitrary commands with elevated privileges. This could potentially lead to a full system compromise. Tracked as CVE-2026-9862 and assigned a CVSS v3.1 score of 9.8, the flaw exists in the boks_autoregisterd service,…
-
Critical LiteLLM Flaw Allows Authentication Bypass via Host Header Injection
Tags: advisory, attack, authentication, cve, cyber, flaw, framework, github, injection, vulnerabilityA critical security vulnerability tracked as CVE-2026-49468 has been disclosed in the LiteLLM framework, exposing deployments to authentication bypass attacks via Host header injection. The issue, published in the GitHub Advisory Database and classified under GHSA-4xpc-pv4p-pm3w, affects all LiteLLM versions before 1.84.0 and has been assigned a critical severity rating due to its potential impact…
-
Critical LiteLLM Flaw Allows Authentication Bypass via Host Header Injection
Tags: advisory, attack, authentication, cve, cyber, flaw, framework, github, injection, vulnerabilityA critical security vulnerability tracked as CVE-2026-49468 has been disclosed in the LiteLLM framework, exposing deployments to authentication bypass attacks via Host header injection. The issue, published in the GitHub Advisory Database and classified under GHSA-4xpc-pv4p-pm3w, affects all LiteLLM versions before 1.84.0 and has been assigned a critical severity rating due to its potential impact…
-
Copilot ‘SearchLeak’ Attack Allows 1-Click Data Theft
The critical, three-stage attack is now patched, but it’s part of a new group of AI prompt-injection issues that use hidden URLs and other variables. First seen on darkreading.com Jump to article: www.darkreading.com/application-security/copilot-searchleak-attack-1-click-data-theft
-
Openclaw durch Prompt-Injection in Nachrichtenobjekten angreifbar
Die Sicherheitsforscher von Thales haben Sicherheitslücken auf Basis von Prompt-Injection bei Openclaw gefunden. Diese Schwachstellen wurden dem Openclaw-Sicherheitsteam verantwortungsbewusst gemeldet und mit Version 2026.4.23 wurde ein entsprechender Fix bereitgestellt. Dennoch bleiben zwei Herausforderungen bestehen: Prompt-Injection ist ein branchenweit weitgehend ungelöstes Problem. Es gibt keinen Standard, der regelt, wie Messaging-Objekte serialisiert werden, bevor sie ein LLM…
-
Palo Alto PAN-OS Flaw Lets Attackers Run Arbitrary Commands With Root Privileges
Palo Alto Networks has released patches for three new PAN-OS vulnerabilities that could allow authenticated administrators or users to execute arbitrary commands with root privileges or force firewalls into repeated reboots, raising operational and security concerns for enterprises relying on PA-Series and VM-Series appliances. PAN-OS Root Command Injection via CLI and Web UI (CVE-2026-0273) CVE-2026-0273…
-
LangGraph Flaw Chain Exposes Self-Hosted AI Agents to Remote Code Execution
Cybersecurity researchers have disclosed details of three now-patched security flaws impacting LangGraph, including a critical vulnerability chain that could result in remote code execution.LangGraph is an open-source framework created by LangChain to build complex, stateful, and multi-agent artificial intelligence (AI) agentic applications.”An SQL injection in LangGraph’s function could First seen on thehackernews.com Jump to article:…
-
OpenClaw ist durch Prompt Injections in Nachrichtenobjekten angreifbar
Sicherheitsteams sollten ihre Sandbox-Isolation aktivieren und das Least-Privilege-Prinzip auf die Berechtigungen der Agenten-Tools anwenden. First seen on infopoint-security.de Jump to article: www.infopoint-security.de/openclaw-ist-durch-prompt-injections-in-nachrichtenobjekten-angreifbar/a45460/
-
KI-Governance im Fokus: Was der Lockdown Mode von OpenAI wirklich sagt
OpenAI hat aktuell den »Lockdown Mode« für ChatGPT angekündigt. Dabei tat das Unternehmen etwas Bemerkenswertes: Es bestätigte öffentlich, dass Prompt Injection über MCP-Konnektoren ein ernstes unternehmerisches Exfiltrationsrisiko darstellt. Es ist ernst genug, um darauf architektonisch zu reagieren. Für Sicherheits- und Compliance-Verantwortliche in deutschen Unternehmen, ob Mittelstand oder DAX-Konzern, ist diese Bestätigung wichtig und hat direkte……
-
CVE-2026-10520 Exploited: Ivanti Sentry Gateways Compromised Shortly After Patch Release
Tags: cve, data-breach, exploit, flaw, injection, Internet, ivanti, remote-code-execution, threat, updateAttackers are exploiting the critical CVE-2026-10520 flaw in Ivanti Sentry, compromising many internet-exposed gateways shortly after patches were released. Threat actors have started exploiting a maximum-severity OS command injection flaw in Ivanti Sentry, tracked as CVE-2026-10520, that allows remote code execution with root privileges. >>An OS Command Injection vulnerability in Ivanti Sentry before the R10.5.2, R10.6.2 and R10.7.1 versions allows a remote…
-
CVE-2026-10520 Exploited: Ivanti Sentry Gateways Compromised Shortly After Patch Release
Tags: cve, data-breach, exploit, flaw, injection, Internet, ivanti, remote-code-execution, threat, updateAttackers are exploiting the critical CVE-2026-10520 flaw in Ivanti Sentry, compromising many internet-exposed gateways shortly after patches were released. Threat actors have started exploiting a maximum-severity OS command injection flaw in Ivanti Sentry, tracked as CVE-2026-10520, that allows remote code execution with root privileges. >>An OS Command Injection vulnerability in Ivanti Sentry before the R10.5.2, R10.6.2 and R10.7.1 versions allows a remote…
-
Fortinet patched a new critical FortiSandbox flaw
Fortinet patched a critical FortiSandbox vulnerability that could let unauthenticated attackers remotely execute commands via crafted HTTP requests. Fortinet released security updates to address several vulnerabilities affecting FortiSandbox, FortiOS, FortiProxy, and FortiPortal. The most severe issue, tracked as CVE-2026-25089 (CVSS score of 9.8), is an OS command injection flaw in FortiSandbox products. The vulnerability could…
-
Ivanti Command Injection Flaw Exploited After PoC Code Release
Ivanti Sentry is facing active exploitation attempts following the public release of proof-of-concept (PoC) code targeting a critical OS command injection vulnerability tracked as CVE-2026-10520. The flaw, along with a second critical issue (CVE-2026-10523), was disclosed by Ivanti on June 9, 2026, with both affecting multiple versions of the Sentry mobile device management gateway. Although…
-
Prompt injection still drives most agentic AI security failures in production
A backdoor sat on PyPI for three hours in March 2026. Nearly 47,000 downloads occurred during the window. The compromised package, LiteLLM, serves as the language-model … First seen on helpnetsecurity.com Jump to article: www.helpnetsecurity.com/2026/06/11/owasp-prompt-injection-ai-security-failures/
-
LiteLLM vulnerability under active attack, CISA warns (CVE-2026-42271)
Tags: ai, attack, cisa, cybersecurity, exploit, infrastructure, injection, open-source, vulnerabilityA command injection vulnerability (CVE-2026-42271) in BerryAI’s LiteLLM open-source AI gateway is being exploited by attackers, the US Cybersecurity and Infrastructure … First seen on helpnetsecurity.com Jump to article: www.helpnetsecurity.com/2026/06/09/litellm-vulnerability-under-active-attack-cisa-warns-cve-2026-42271/
-
LiteLLM Flaw CVE-2026-42271 Exploited in the Wild, Chains to Unauthenticated RCE
Tags: cve, cybersecurity, exploit, flaw, infrastructure, injection, kev, rce, remote-code-execution, vulnerabilityThe U.S. Cybersecurity and Infrastructure Security Agency (CISA) on Monday added a high-severity flaw impacting BerriAI LiteLLM to its Known Exploited Vulnerabilities (KEV) catalog, citing evidence of active exploitation.The vulnerability, tracked as CVE-2026-42271 (CVSS score: 8.7), is a command injection vulnerability that could allow any authenticated user to run arbitrary commands on the First seen…
-
OpenAI Expands ChatGPT Lockdown Mode to Millions of Eligible Users
OpenAI is expanding ChatGPT Lockdown Mode to more users, limiting web-connected tools to reduce the risks of prompt injection and data leakage. The post OpenAI Expands ChatGPT Lockdown Mode to Millions of Eligible Users appeared first on TechRepublic. First seen on techrepublic.com Jump to article: www.techrepublic.com/article/news-openai-expands-chatgpt-lockdown-mode-millions-users/
-
OpenAI Unveils ChatGPT Account Security Controls
OpenAI brings Lockdown Mode and Active Sessions to ChatGPT to curb prompt injection data theft First seen on infosecurity-magazine.com Jump to article: www.infosecurity-magazine.com/news/chatgpt-lockdown-mode-active/
-
Infosecurity Europe: Prompt Injection Remains Unsolved, OWASP Researcher Warns
At Infosecurity Europe 2026, OWASP’s Ariel Fogel warned that prompt injection remains an “unresolved problem” within generative AI architecture First seen on infosecurity-magazine.com Jump to article: www.infosecurity-magazine.com/news/infosec-europe-prompt-injection/
-
Critical UniFi OS RCE Chain Grants Root Access Without Credentials
Tags: access, advisory, authentication, credentials, cyber, flaw, injection, rce, remote-code-execution, update, vulnerabilitySecurity Advisory Bulletin 064 describing a critical chain of vulnerabilities in UniFi OS Server that allows unauthenticated remote code execution and full root takeover. The issue combines an authentication-gateway bypass, a path-traversal mismatch, and a command-injection sink in the package-update service. When chained, these flaws let an attacker send a single crafted HTTP request to…
-
EDRChoker Tool Abuses Windows QoS Policies to Disrupt Endpoint Security Tools
A newly disclosed red-team tool dubbed “EDRChoker” is drawing attention across the cybersecurity community for its novel approach to disrupting Endpoint Detection and Response (EDR) visibility by abusing Windows Policy-based Quality of Service (quality of service). Unlike traditional EDR evasion techniques that rely on firewall manipulation or Windows Filtering Platform (WFP) rule injection, EDRChoker operates…
-
New ChatGPT Lockdown Mode Aims to Block Prompt Injection and Data Exfiltration Attacks
OpenAI this week introduced Lockdown Mode, a security-focused setting for ChatGPT designed to reduce the risk of data exfiltration from prompt-injection attacks. The feature is rolling out to eligible personal accounts (Free, Go, Plus, Pro) and self-serve ChatGPT Business workspaces, and managed-workspace administrators can assign a Lockdown Mode role to members. Prompt injection is a…
-
OpenAI unveils Lockdown Mode to protect sensitive data from prompt injection attacks
Even with Lockdown Mode, ChatGPT could be still vulnerable to prompt injections, but the goal is to reduce the likelihood that sensitive data gets shared in the process. First seen on techcrunch.com Jump to article: techcrunch.com/2026/06/06/openai-unveils-lockdown-mode-to-protect-sensitive-data-from-prompt-injection-attacks/
-
New ChatGPT Lockdown Mode Limits Tools That Could Enable Data Exfiltration
OpenAI has begun rolling out a new Lockdown Mode to ChatGPT for eligible personal accounts to reduce the risk of data exfiltration arising from prompt injection attacks.The feature is primarily designed for people and organizations that handle sensitive data and require stricter protection guarantees. Lockdown Mode is available to logged-in users across Free, Go, Plus,…
-
Android Gemini prompt injection flaw patched by Google
First seen on scworld.com Jump to article: www.scworld.com/brief/android-gemini-prompt-injection-flaw-patched-by-google
-
Cisco warns zero-day flaw in SD-WAN is being exploited
The company cautioned that no current patches are available and the flaw could allow an attacker to conduct command injection attacks. First seen on cybersecuritydive.com Jump to article: www.cybersecuritydive.com/news/cisco-zero-day-flaw-sd-wan-exploited/822138/
-
Null-Byte umgeht Netzwerk-Allowlist in Claude Code – Sandbox-Bypass öffnet Claude Code für PromptAngriffe
First seen on security-insider.de Jump to article: www.security-insider.de/claude-code-sandbox-bypass-null-byte-prompt-injection-datenabfluss-a-f4dc130466f4aed616ca8b5d64613fc5/
-
Critical Apache ActiveMQ Vulnerability Exposes Systems to Security Header Injection Attacks
Apache ActiveMQ users are being urged to apply immediate patches following the disclosure of a critical vulnerability, CVE-2026-42253, that enables HTTP response header injection via improperly handled JMS message properties. The flaw affects both Apache ActiveMQ and ActiveMQ Web components. It has been rated with “important” severity by the Apache Software Foundation. CVE-2026-42253: HTTP Response…