Tag: injection
-
MCP: securing the backbone of Agentic AI
Tags: access, ai, attack, authentication, business, ciso, control, credentials, cyber, data, detection, injection, least-privilege, mfa, monitoring, RedTeam, risk, security-incident, service, supply-chain, trainingFour cornerstones for securing MCP servers: CISOs can largely rely on the proven basic principles of cyber security for MCP they just need to adapt them in a few places. Pure checklists fall short here. Instead, a clear, principles-based approach is required. Four central pillars have proven themselves in practice: Strong authentication and clean credential…
-
BSidesSF 2025: GenAI Application Security: Not Just Prompt Injection
Creator/Author/Presenter: Ahmed Abugharbia Our deep appreciation to Security BSides – San Francisco and the Creators/Authors/Presenters for publishing their BSidesSF 2025 video content on YouTube. Originating from the conference’s events held at the lauded CityView / AMC Metreon – certainly a venue like no other; and via the organization’s YouTube channel. Additionally, the organization is welcoming…
-
Cursor AI Code Editor Fixed Flaw Allowing Attackers to Run Commands via Prompt Injection
Cybersecurity researchers have disclosed a now-patched, high-severity security flaw in Cursor, a popular artificial intelligence (AI) code editor, that could result in remote code execution.The vulnerability, tracked as CVE-2025-54135 (CVSS score: 8.6), has been addressed in version 1.3 released on July 29, 2025. It has been codenamed CurXecute by Aim”¯Labs, which previously disclosed EchoLeak. First…
-
AI-powered Cursor IDE vulnerable to prompt-injection attacks
A vulnerability that researchers call CurXecute is present in almost all versions of the AI-powered code editor Cursor, and can be exploited to execute remote code with developer privileges. First seen on bleepingcomputer.com Jump to article: www.bleepingcomputer.com/news/security/ai-powered-cursor-ide-vulnerable-to-prompt-injection-attacks/
-
How bright are AI agents? Not very, recent reports suggest
CSOs should ‘skip the fluff’: Meghu’s advice to CSOs: Stop reading the marketing and betting too much of your business on AI/LLM technology as it exists today. Start small and always have a human operator to guide it.”If you skip the fluff and get to the practical application, we have a new technology that could…
-
What Amazon Q prompt injection reveals about AI security
Experts say a malicious prompt injection in the Amazon Q extension for VS Code doesn’t represent a fundamentally new threat, but reflects how AI amplifies security risks. First seen on techtarget.com Jump to article: www.techtarget.com/searchsoftwarequality/news/366628167/What-Amazon-Q-prompt-injection-reveals-about-AI-security
-
Securing the Next Era: Why Agentic AI Demands a New Approach to API Security
I’ve spent my career building solutions to protect the API fabric that powers modern businesses. I founded Salt because I saw that traditional security tools such as WAFs, gateways, and CDNs weren’t designed to see or secure APIs. That gap led to breaches, blind spots, and billions in risk. Today, we’re facing a new wave…
-
Enterprise LLMs Vulnerable to Prompt-Based Attacks Leading to Data Breaches
Security researchers have discovered alarming vulnerabilities in enterprise Large Language Model (LLM) applications that could allow attackers to bypass authentication systems and access sensitive corporate data through sophisticated prompt injection techniques. The findings reveal that many organizations deploying AI-powered chatbots and automated systems may be inadvertently exposing critical information to malicious actors. The vulnerability stems…
-
New Microsoft Guidance Targets Defense Against Indirect Prompt Injection
Microsoft has unveiled new guidance addressing one of the most pressing security challenges facing enterprise AI deployments: indirect prompt injection attacks. This emerging threat vector has become the top entry in the OWASP Top 10 for LLM Applications & Generative AI 2025, prompting the tech giant to develop a multi-layered defense strategy spanning prevention, detection,…
-
Google patches Gemini CLI tool after prompt injection flaw uncovered
README.md GNU Public License file of the sort that would be part of any open source repo.The researchers then uncovered a combination of smaller weaknesses that could be exploited together to run malicious shell commands without the user’s knowledge. The first weakness is that Gemini CLI sensibly allows users to allowlist frequent commands, for example,…
-
Gemini CLI Vulnerability Allows Silent Execution of Malicious Commands on Developer Systems
Security researchers at Tracebit have discovered a critical vulnerability in Google’s Gemini CLI that enables attackers to silently execute malicious commands on developers’ systems through a sophisticated combination of prompt injection, improper validation, and misleading user interface design. The vulnerability, classified as a P1/S1 issue by Google’s security team, has been patched in the latest…
-
Why React Didn’t Kill XSS: The New JavaScript Injection Playbook
React conquered XSS? Think again. That’s the reality facing JavaScript developers in 2025, where attackers have quietly evolved their injection techniques to exploit everything from prototype pollution to AI-generated code, bypassing the very frameworks designed to keep applications secure.Full 47-page guide with framework-specific defenses (PDF, free).JavaScript conquered the web, but with First seen on thehackernews.com…
-
CISA Issues Alert on Cisco Identity Services Engine Flaw Exploited in Active Attacks
Tags: attack, cisa, cisco, cyber, cybersecurity, exploit, flaw, identity, infrastructure, injection, kev, risk, service, threat, vulnerabilityThe Cybersecurity and Infrastructure Security Agency (CISA) has issued a critical security alert regarding severe vulnerabilities in Cisco’s Identity Services Engine (ISE) that are being actively exploited by threat actors. The agency added two critical injection vulnerabilities to its Known Exploited Vulnerabilities Catalog on July 28, 2025, signaling immediate risks to organizations using the affected…
-
Entwickler-Tool von Amazon verseucht
Tags: access, ai, cloud, cyberattack, cybersecurity, github, governance, hacker, injection, monitoring, open-source, risk, supply-chain, tool, update, vulnerabilityAuch die leistungsstärksten KI-Tools sind kontraproduktiv, wenn sie nicht richtig abgesichert sind. Einem Hacker ist es gelungen, zerstörerische Systembefehle in die Visual-Studio-Code-Extension einzuschleusen, die für den Zugriff auf Amazons KI-gestützten Programmierassistenten Q verwendet wird. Der Angreifer konnte das Entwickler-Tool (mit mehr als 950.000 Installationen) über ein nicht-verifiziertes GitHub-Konto verseuchen: Er reichte Ende Juni 2025 einen…
-
Hackers Use Weaponized .HTA Files to Infect Victims with Red Ransomware
CloudSEK’s TRIAD team uncovered an active development site deploying Clickfix-themed malware linked to the Epsilon Red ransomware. This variant deviates from traditional clipboard-based command injection tactics by directing victims to a secondary page on the same domain, where malicious shell commands are executed silently through ActiveXObject(>>WScript.Shell
-
Neuen Bedrohungskategorie – Check Point entdeckt Prompt-Injection-Versuch
Tags: injectionFirst seen on security-insider.de Jump to article: www.security-insider.de/neue-malware-ki-manipulation-prompt-injection-a-e6b5d4e8af11081a484229b103216641/
-
Microsoft ‘digital escorts’ reveal crucial US counterintelligence blind spot
Tags: access, china, cio, cloud, compliance, country, cyber, cybersecurity, data, defense, firewall, framework, google, government, injection, intelligence, law, microsoft, military, oracle, risk, service, threat, update, vulnerabilityWhat the program was, and how it worked: The digital escort model, according to ProPublica, was designed to comply with federal contracting rules that prohibit foreign nationals from directly accessing sensitive government systems. Under this framework:China-based engineers would file support tickets for tasks such as firewall updates or bug fixes.US-based escorts, often former military personnel…
-
Week in review: Google fixes zero-day vulnerability in Chrome, critical SQL injection flaw in FortiWeb
Here’s an overview of some of last week’s most interesting news, articles, interviews and videos: Update Google Chrome to fix actively exploited zero-day (CVE-2025-6558) For … First seen on helpnetsecurity.com Jump to article: www.helpnetsecurity.com/2025/07/20/week-in-review-google-fixes-zero-day-vulnerability-in-chrome-critical-sql-injection-flaw-in-fortiweb/
-
Google Gemini AI Flaw Could Lead to Gmail Compromise, Phishing
Researchers discovered a security flaw in Google’s Gemini AI chatbot that could put the 2 billion Gmail users in danger of being victims of an indirect prompt injection attack, which could lead to credentials being stolen or phishing attacks. First seen on securityboulevard.com Jump to article: securityboulevard.com/2025/07/google-gemini-ai-flaw-could-lead-to-gmail-compromise-phishing/
-
Hackers Are Finding New Ways to Hide Malware in DNS Records
Newly published research shows that the domain name system”, a fundamental part of the web”, can be exploited to hide malicious code and prompt injection attacks against chatbots. First seen on wired.com Jump to article: www.wired.com/story/dns-records-hidden-malicious-code/
-
Nvidia-GPUs anfällig für Rowhammer-Angriffe
Cyberkriminelle könnten über einen Rowhammer-Angriff auf Nvidia-GPUs zugreifen.Nvidia hat kürzlich eine Sicherheitswarnung an Anwendungsentwickler, Computerhersteller und IT-Verantwortliche herausgegeben, dass moderne Speicherchips in Grafikprozessoren potenziell anfällig für sogenannte Rowhammer-Exploits sind. Bereits zuvor hatten kanadische Universitätsforscher nachgewiesen, dass eine Nvidia A6000 GPU mit einem ähnlichen Angriff erfolgreich kompromittiert werden kann.Bei einem Rowhammer-Angriff handelt es sich um eine…
-
LaRecipe Tool with 2.3M Downloads Found Vulnerable to Full Server Takeover
A critical security vulnerability has been discovered in LaRecipe, a popular Laravel documentation package with over 2.3 million downloads, that could allow attackers to completely compromise affected servers. The vulnerability, identified as CVE-2025-53833, enables Server-Side Template Injection (SSTI) attacks that can lead to Remote Code Execution (RCE) on vulnerable systems. Critical Vulnerability Discovered Security researcher…
-
AI poisoning and the CISO’s crisis of trust
Tags: access, ai, breach, ceo, ciso, compliance, control, cybersecurity, data, defense, detection, disinformation, exploit, framework, healthcare, identity, infosec, injection, LLM, monitoring, network, privacy, RedTeam, resilience, risk, russia, saas, threat, tool, trainingFoundation models began parroting Kremlin-aligned propaganda after ingesting material seeded by a large-scale Russian network known as the “Pravda Network.”A high-profile AI-generated reading list published by two American news outlets included 10 hallucinated book titles mistakenly attributed to real authors.Researchers showed that imperceptible perturbations in training images could trigger misclassification. Researchers in the healthcare domain demonstrated…
-
Wing FTP Vulnerability Actively Exploited Globally
Remote Code Execution Flaw Affects More Than 5,000 Servers. Threat actors are actively exploiting a critical vulnerability in a server file transfer solution. Researchers say the flaw in Wing FTP Server could allow threat actors to execute system-level commands remotely, using null byte and Lua injection without authentication. First seen on govinfosecurity.com Jump to article:…
-
Google Gemini AI Bug Allows Invisible, Malicious Prompts
A prompt-injection vulnerability in the AI assistant allows attackers to create messages that appear to be legitimate Google Security alerts but instead can be used to target users across various Google products with vishing and phishing. First seen on darkreading.com Jump to article: www.darkreading.com/remote-workforce/google-gemini-ai-bug-invisible-malicious-prompts
-
Summarizing Emails With Gemini? Beware Prompt Injection Risk
Attackers Can Trick Gemini Into Displaying Deceptive Messages, Researchers Warn. Attackers can hide malicious instructions inside emails to trick Google’s Gemini into delivering falsified summaries with deceptive messages to end users, researchers warn. Google said it’s continuing to put multiple defenses in place to combat these types of prompt injection attacks. First seen on govinfosecurity.com…
-
Exploits for unauthenticated FortiWeb RCE are public, so patch quickly! (CVE-2025-25257)
With two proof-of-concept (PoC) exploits made public late last week, CVE-2025-25257 a critical SQL command injection vulnerability in Fortinet’s FortiWeb web … First seen on helpnetsecurity.com Jump to article: www.helpnetsecurity.com/2025/07/14/exploits-for-unauthenticated-fortiweb-rce-are-public-so-patch-quickly-cve-2025-25257/