Tag: injection
-
CrowdStrike Uncovers 5 New Prompt Injection Techniques Targeting AI Agents
CrowdStrike has identified five new techniques for prompt injection targeting AI agents, emphasizing the rapid evolution of adversarial methods as enterprises increasingly deploy autonomous AI systems. Detailed in a report published on July 7, 2026, by CrowdStrike’s AI security research team, these techniques expand the company’s prompt injection taxonomy to over 200 documented attack methods.…
-
Ubiquiti warns of new max severity UniFi OS vulnerability
Ubiquiti has released security updates to patch seven critical vulnerabilities in UniFi OS, including a maximum-severity flaw that can be exploited in command injection attacks. First seen on bleepingcomputer.com Jump to article: www.bleepingcomputer.com/news/security/ubiquiti-warns-of-new-max-severity-unifi-os-vulnerability/
-
GitLost: GitHub’s AI Agent Tricked Into Leaking Private Repository Data
Noma Labs details GitLost, a prompt injection flaw that made GitHub’s AI agent expose private repo data through a crafted public issue and guardrail failures. First seen on hackread.com Jump to article: hackread.com/gitlost-github-ai-agent-leaking-repository-data/
-
Thousands of MCP Servers Found Vulnerable to File Access and Injection Attacks
Thousands of Model Context Protocol (MCP) servers, widely used to connect large language models (LLMs) to external systems, have been found vulnerable to critical security flaws, including arbitrary file access, command injection, server-side request forgery (SSRF), and SQL injection, raising significant concerns about AI supply chain security. A large-scale analysis of 9,695 MCP servers across…
-
Indirect Prompt Injection in Web Content Targets AI Agents
Zscaler found sites hiding prompt-injection text to manipulate AI agents into crypto payments First seen on infosecurity-magazine.com Jump to article: www.infosecurity-magazine.com/news/indirect-prompt-injection-web/
-
Hidden Web Prompts Trick AI Agents Into Sending Money
Hidden prompts on malicious websites trick AI agents into making payments or trusting fake sites, exposing new risks for autonomous AI workflows. Zscaler ThreatLabz documented two active campaigns that embed hidden instructions in web pages to manipulate AI agents, not human users, though those get caught too. The technique is called indirect prompt injection: malicious…
-
KI-Exploit durch Prompt Injection – Cursor-Schwachstellen hebeln Sandbox aus
First seen on security-insider.de Jump to article: www.security-insider.de/cursor-ide-duneslide-sicherheitsluecken-prompt-injection-sandbox-bypass-a-92a2a3d25ac1c9726ae9afde76020cd1/
-
Hackers Use Fake API Documentation to Trick AI Agents Into Sending Crypto Payments
Hackers are now weaponizing documentation and site metadata to mislead autonomous AI agents into executing cryptocurrency payments. The attack leverages indirect prompt injection (IPI): malicious instructions hidden in web content and structured data that influence an AI agent’s reasoning during automated tasks. By combining SEO poisoning, JSON”‘LD abuse and CSS concealment, attackers create seemingly legitimate…
-
BioShocking-Angriff hebelt KI-Schutzfilter aus
Der Prompt-Injection-Angriff BioShocking bringt KI-Browser dazu, Sicherheitsfilter zu ignorieren und sensible Nutzerdaten wie Passwörter zu stehlen. First seen on it-daily.net Jump to article: www.it-daily.net/it-sicherheit/cybercrime/bioshocking-angriff-ki-schutzfilter
-
Opera Browser Adds Native Paste Protect to Stop Clipboard Hijacking and Code Injection Attacks
Opera has announced a new native security feature called “Paste Protect,” which aims to combat clipboard hijacking and command injection attacks directly within the browser. This marks a significant advancement in proactive endpoint protection at the user interaction level. Introduced on July 2, 2026, the feature is enabled by default. It addresses a rapidly growing…
-
ValleyRAT Uses RC4 Encryption, Donut Shellcode, and rundll32 Injection for Stealth
A recent surge in ValleyRAT activity that combines RC4-encrypted payloads, Donut-generated shellcode, and in-memory execution via suspended rundll32 processes to evade detection. First named by Proofpoint in 2023, ValleyRAT continues to evolve: LevelBlue’s telemetry shows a marked increase in successful detections beginning May 2025 and accelerating into 2026. The threat now presents through two primary…
-
Critical Cursor IDE Flaws Let Attackers Execute Code via Zero-Click Prompt Injection
Two significant remote code execution (RCE) vulnerabilities in the widely used Cursor ID expose developers to zero-click attacks driven by prompt injection. These vulnerabilities, tracked as CVE-2026-50548 and CVE-2026-50549, collectively known as >>DuneSlide,<< carry a CVSS score of 9.8. They demonstrate how development environments powered by large language models (LLMs) can unintentionally increase the attack…
-
Critical Cursor Flaws Could Let Prompt Injection Escape Sandbox and Run Commands
Two flaws in Cursor, an AI code editor, could let a single, ordinary-looking prompt break out of the editor’s safety sandbox and run any command on a developer’s computer. There is no click to fall for and no approval box to ignore.Cato AI Labs found the pair and named them DuneSlide. They are tracked as…
-
Progress Kemp LoadMaster Pre-Auth RCE Flaw Faces Active Exploitation Attempts
A recently disclosed critical security flaw impacting Progress”¯Kemp LoadMaster is seeing active exploitation attempts, according to an advisory from eSentire’s Threat Response Unit (TRU).The Canadian cybersecurity company said it identified exploitation attempts targeting CVE-2026-8037 (CVSS score: 9.6), an operating system (OS) command injection flaw that could be exploited to achieve First seen on thehackernews.com Jump…
-
ObjectLücke gefährdet Magento-Shops – CVE-2026-45247 in Mirasvit Full Page Cache Warmer aktiv ausgenutzt
First seen on security-insider.de Jump to article: www.security-insider.de/magento-mirasvit-full-page-cache-warmer-cve-2026-45247-a-02011748fec50683ce828c03a2e793a6/
-
GuardFall Flaw Hits 10 of 11 Popular Open-Source AI Agents
Researchers found a shell injection flaw in 10 of 11 popular open-source AI agents, allowing attackers to bypass command filters. Adversa AI just published a survey, titled >>GuardFall: a universal shell injection vulnerability in open-source AI agents,<< of eleven open-source AI coding and computer-use agents, and the headline finding is uncomfortable: ten of them leave…
-
New BioShocking attack manipulates AI browser into data theft
A new prompt injection attack dubbed “BioShocking” could trick AI-powered browsers into treating real-world risky actions as part of a fictional scenario, causing them to ignore any safety guardrails. First seen on bleepingcomputer.com Jump to article: www.bleepingcomputer.com/news/security/new-bioshocking-attack-manipulates-ai-browser-into-data-theft/
-
GuardFall Exposes Open-Source AI Coding Agents to Decades-Old Shell Injection Risks
The safety check that is supposed to stop an AI coding agent from running a dangerous command can be walked straight past using a shell trick that has been public for decades.New research from Adversa AI, which is named the bypass GuardFall, found it works against ten of the eleven popular open-source coding and computer-use…
-
North Korea-Linked macOS Malware Uses Prompt Injection to Evade AI Analysis
SentinelOne says macOS.Gaslight uses prompt injection to mislead AI-based malware analysis, steal data, and use Telegram for C2. The post North Korea-Linked macOS Malware Uses Prompt Injection to Evade AI Analysis appeared first on TechRepublic. First seen on techrepublic.com Jump to article: www.techrepublic.com/article/news-macos-gaslight-malware-ai-prompt-injection/
-
Mozilla warns of indirect prompt injection risk in AI coding agents
A malicious GitHub repository can silently compromise a developer’s machine without containing a single line of malicious code, security researchers at Mozilla’s … First seen on helpnetsecurity.com Jump to article: www.helpnetsecurity.com/2026/06/29/mozilla-warns-of-indirect-prompt-injection-risk-in-ai-coding-agents/
-
New Windows Injection Technique Hijacks Win32k Callback Dispatch to Execute Shellcode
A newly documented injection technique abuses the kernel-to-user callback dispatch path used by the Windows graphical subsystem (win32k.sys) to achieve remote code execution while leaving the KernelCallbackTable structurally intact. Rather than replacing a KernelCallbackTable entry with a shellcode pointer, the operator resolves a legitimate callback target from the table and installs an inline detour inside…
-
Five Eyes Warns AI Could Speed Cyberattacks Within Months
Five Eyes agencies warned that AI could speed cyberattacks within months, raising new risks around prompt injection, phishing, and enterprise AI tools. The post Five Eyes Warns AI Could Speed Cyberattacks Within Months appeared first on TechRepublic. First seen on techrepublic.com Jump to article: www.techrepublic.com/article/news-five-eyes-ai-cyberattacks/
-
FOSSBilling Flaw Lets Admin Attackers Abuse DI Container for SQL Access and RCE
A critical server-side template injection (SSTI) vulnerability in FOSSBilling, tracked as CVE-2026-28496, is exposing instances to potential full database compromise and remote code execution (RCE), with early signs of active exploitation appearing shortly after public disclosure. This flaw is documented under GitHub advisory GHSA-57mv-jm88-66jc and affects all versions up to 0.7.2. It has been patched…
-
macOS.Gaslight: North Korea-Linked Malware That Tries to Gaslight the Analyst
macOS.Gaslight: DPRK Rust implant for Mac with a prompt injection payload designed to fool AI-based malware analysts. SentinelLabs researchers spotted a Rust-based macOS implant, dubbed macOS.Gaslight, that surfaced in early June after an Apple XProtect update pointed to a VirusTotal sample uploaded on May 22. The binary was undetected by static engines at the time…
-
DPRK-Linked macOS Implant Uses LaunchAgent Persistence and Python Stealer Module
The binary tracked as macOS.Gaslight as a Rust-based macOS implant and infostealer whose most novel features are analyst-directed prompt injection and a hardened Telegram-based command-and-control (C2) channel. We assess with high confidence that macOS.Gaslight aligns with DPRK-linked macOS activity clustered around BONZAI and AIRPIPE signatures. macOS.Gaslight is ad hoc signed, carries the identifier endpoint-macos-aarch64-5555494492fc075f441637fb9d894913dde3a2ea, and…
-
Chrome Ad Blocker with 10M+ Installs Found with Dormant Script Injection Capability
An analysis of a popular Google Chrome ad block extension for YouTube has uncovered the ability to execute arbitrary JavaScript code.According to Island, the extension, named Adblock for YouTube (ID: cmedhionkhpnakcndndgjdbohmhepckk), has more than 10 million installs and carries a Featured badge on the Chrome Web Store.The extension description states that it allows users to…
-
New Gaslight macOS Malware Uses Prompt Injection to Disrupt AI-Assisted Analysis
A previously undocumented Rust-based macOS implant and information stealer has been found to embed a prompt injection payload designed to trick a malware analyst’s artificial intelligence (AI) tools and trick it into aborting or refusing an analysis of the artifact.The malware has been codenamed Gaslight owing to this deceptive behavior. It’s been assessed with high…
-
CISA Warns Critical Lantronix EDS5000 Flaw Is Being Actively Exploited
The U.S. Cybersecurity and Infrastructure Security Agency (CISA) on Tuesday warned of active exploitation of a critical security flaw impacting Lantronix EDS5000 Series devices, urging Federal Civilian Executive Branch (FCEB) agencies to apply the fixes by June 26, 2026.The vulnerability in question is CVE-2025-67038 (CVSS score: 9.8), a code injection flaw that could result in…
-
macOS Backdoor Uses Prompt Injection to Evade AI Triage
SentinelLabs found a North Korea-linked macOS backdoor using prompt injection on AI triage tools First seen on infosecurity-magazine.com Jump to article: www.infosecurity-magazine.com/news/macos-gaslight-rust-backdoor/