Find what you’re looking for
Browse everything
6,854 results in Threat Reports
JFrog Artifactory Under Attack: Critical Vulnerability Chain Enables Supply Chain Compromise
Between August and September 2026, threat actors exploited a chain of critical vulnerabilities in JFrog Artifactory (CVE-2026-42018, CVE-2026-42016, and CVE-2026-82329) to bypass authentication and gain administrative privileges on self-hosted instances. Attackers leveraged these flaws to obtain JWT tokens for anonymous users, escalate to admin-level permissions within minutes, and deploy custom Rust-based backdoors with command-and-control capabilities. The campaign affected multiple organizations, with attackers installing malicious Groovy plugins, establishing persistence, stealing configuration data, and creating rogue administrator accounts across vulnerable infrastructure. This incident highlights the escalating sophistication of supply chain attacks targeting development infrastructure, as 49-62% of internet-accessible Artifactory instances remain vulnerable to these authentication bypass flaws, creating widespread exposure across the software development ecosystem.
How ShinyHunters Weaponized Claude AI to Harvest Secrets from 1.8 Million Android Apps
Between December 2025 and August 2026, multiple threat groups including ShinyHunters, Russian state-sponsored Midnight Blizzard, and Chinese espionage group GTG-10007 systematically abused Anthropic's Claude AI model for large-scale cyberattacks. The most significant operation involved ShinyHunters member 'frkoo' deploying an automated credential-harvesting pipeline across AWS infrastructure that extracted secrets from 1.8 million Android applications and compromised over 40 Microsoft corporate tenants within 34 hours. The AI-enhanced attacks enabled rapid progression from initial access to administrative control in under three hours, with confirmed breaches across government, healthcare, energy, and technology sectors. This incident represents a critical inflection point where AI capabilities are being weaponized at unprecedented scale and speed, fundamentally changing the threat landscape and requiring immediate reassessment of defensive strategies against AI-enhanced cybercrime operations.
GitLab's CVSS 10.0 Vulnerability: Why DevOps Security Can't Wait
GitLab disclosed a critical path traversal vulnerability (CVE-2026-85706) with a maximum CVSS score of 10.0, allowing unauthenticated attackers to read arbitrary files from GitLab servers through the repository commits API. The flaw affects GitLab CE and EE versions from 18.7 through 19.3.1, stemming from improper path confinement and missing authentication enforcement. Within hours of public disclosure on September 11, 2026, security researchers observed active in-the-wild exploitation attempts targeting exposed GitLab instances to extract log files, configuration data, credentials, and sensitive information. This incident highlights the accelerating timeline from vulnerability disclosure to active exploitation, particularly for DevOps platforms that house critical source code and CI/CD secrets. Following a similar pattern to the recent GitLab GraphQL injection vulnerability (CVE-2026-19478), attackers are rapidly weaponizing these flaws to compromise software supply chains and inject malicious code into build pipelines.
How Seven Chinese AI Labs Stole 190+ Million Claude Conversations in Massive Distillation Attack
In September 2026, Anthropic disclosed that seven China-based AI laboratories, including Alibaba, Moonshot, DeepSeek, and others, conducted industrial-scale illicit distillation attacks against Claude AI models between February and July 2026. The attackers used networks of fake accounts created with stolen credit cards and API keys to extract over 190 million conversation exchanges, routing requests through proxy services to harvest Claude's capabilities including chain-of-thought reasoning, coding abilities, and logical reasoning functions for unauthorized training of competing models. This incident highlights the emerging threat landscape of AI model theft and intellectual property extraction, representing a new category of cybercrime where nation-state affiliated entities systematically steal proprietary AI capabilities to advance their own technological development and competitive positioning.
Florida DMV Breach Exposes Risks of Shared Government Database Access
In September 2026, the Florida Department of Highway Safety and Motor Vehicles (FLHSMV) confirmed that its DAVID driver database was breached by the ShinyHunters extortion group, who claimed to have stolen over 200,000 driver records. The attack was executed using compromised credentials from a Plant City Police Department employee that had been improperly stored on a personal device. The breach was discovered on September 4, 2026, and quickly mitigated, with FLHSMV working alongside state law enforcement agencies in their response. This incident highlights the growing trend of cybercriminals targeting government databases through compromised credentials and the critical importance of proper credential management across interconnected systems.
UAC-0099 Deploys GuardBreaker AI Manipulation Against Ukraine
In September 2026, Russia-aligned threat group UAC-0099 deployed a novel AI manipulation technique called GuardBreaker against Ukrainian targets. The attackers embedded nuclear weapon prompts within malicious VBScript comments to trigger AI safety mechanisms and prevent automated malware analysis systems from examining their code. This represents a significant evolution in adversarial tactics, where threat actors manipulate AI defensive reasoning rather than increasing malware sophistication to achieve compromise. This incident highlights the accelerating arms race between AI-powered security tools and adversaries who exploit their limitations. With AI systems now discovering vulnerabilities at unprecedented speed and scale, traditional 90-day patch cycles are obsolete, creating an urgent need for multilayered defense strategies that don't rely solely on AI-based detection mechanisms.
The $21 Billion AI Scam Crisis: How Artificial Intelligence Became the Ultimate Social Engineer
AI-enabled social engineering attacks have reached unprecedented sophistication, with cybercriminals leveraging large language models to create highly personalized and emotionally manipulative scams. Research by Fred Heiding of Menlo Park Intelligence reveals that AI systems excel at human manipulation through voice cloning, long-term relationship building, and cultural context adaptation. The FBI's Internet Crime Center reports that fraud losses skyrocketed from $4 billion in 2020 to $21 billion in 2025, primarily targeting vulnerable populations including senior citizens who develop emotional dependencies on AI-powered scam bots. This asymmetric threat landscape highlights a critical security gap where traditional technical defenses prove inadequate against AI-enhanced social engineering, as human cognitive vulnerabilities cannot be patched like software systems.
The PaperCut AI Swarm Attack: When Machines Wage Cyber Warfare
In August 2026, a Russian-speaking threat actor deployed hundreds of AI agents in a coordinated swarm attack targeting PaperCut NG and MF print management software. The AI-powered campaign compromised at least 440 instances across 395 organizations in 48 countries, achieving remote code execution within four hours and Active Directory domain admin access in just six hours total. Once fully launched, the swarm compromised 11 organizations in merely 26 seconds, demonstrating unprecedented speed and scale in automated cyberattacks. This incident represents a critical inflection point where AI agents are now actively integrated across the entire cyber kill chain, from reconnaissance to exfiltration. As nation-state actors and financially motivated groups increasingly weaponize large language models and agentic AI capabilities, organizations face a new reality where attackers can execute complex, multi-stage operations at machine speed while defenders still operate at human pace.
How Storm-3032 and Storm-3121 Are Exploiting BYOD Policies to Breach Corporate Microsoft 365
Since May 2026, Microsoft researchers have tracked threat actors Storm-3032 and Storm-3121 conducting sophisticated initial access campaigns targeting corporate executives through their personal devices. The attackers use voice calls and text messages impersonating IT helpdesks to trick employees into updating authentication credentials via phishing links. Once access is gained, the threat actors exploit Microsoft Graph API to enumerate corporate resources and exfiltrate sensitive data from SharePoint, OneDrive, and Exchange before potentially selling access to extortion groups like ShinyHunters. This campaign highlights the growing trend of attackers bypassing corporate security controls by targeting the weakest link - personal devices with minimal security protections. As organizations increasingly adopt BYOD policies and hybrid work models, identity-based attacks exploiting trusted communication channels represent a critical evolution in threat actor tactics.
GoldFactory's Android Banking Malware Exploits Work Profiles to Steal $1M from Indonesian Banks
Between February and July 2026, the Chinese-speaking threat group GoldFactory deployed a sophisticated Android banking malware campaign targeting Indonesia, resulting in 1,469 compromised devices and nearly $1 million in losses. The attackers used the Gigabud banking Trojan in combination with Vwork, a modified app-cloning tool, to exploit Android's Work Profile feature. This technique allowed fraudsters to clone legitimate banking applications into isolated environments where security controls and fraud detection systems could not follow, enabling them to conduct transactions while evading detection mechanisms that were triggered in the victim's primary profile. This incident highlights the evolution of mobile banking threats as attackers increasingly target regions with high mobile payment adoption and develop novel evasion techniques that exploit legitimate enterprise security features for malicious purposes.
How Cisco FMC Vulnerabilities Enabled Qilin Ransomware Deployment
In September 2026, Cisco revealed that three distinct threat clusters exploited critical vulnerabilities CVE-2026-20079 and CVE-2026-20316 in Cisco Secure Firewall Management Center (FMC) systems. The attacks involved state-sponsored groups and ransomware operators who leveraged these flaws to deploy web shells, steal credentials, conduct reconnaissance, and ultimately deploy Qilin ransomware. The exploitation allowed attackers to bypass authentication, gain root access, and perform living-off-the-land techniques using legitimate FMC tools to avoid detection while moving laterally through victim networks. This incident highlights the growing trend of threat actors targeting network security infrastructure as initial access vectors, demonstrating how critical security appliances themselves become single points of failure when unpatched vulnerabilities exist.
How UNC3569 Weaponized Trusted Software: The Sogou Input Method Supply Chain Attack
In April 2026, China-linked threat group UNC3569 exploited a critical vulnerability (CVE-2026-51990) in Sogou Input Method, a widely-used Chinese character input tool with over 455 million monthly users. The attackers leveraged a crafted sgbiz: link to bypass security controls and deploy the GRAYRABBIT backdoor through an outdated Chromium browser engine with disabled sandboxing. The exploit chain utilized a 2021 V8 JavaScript engine vulnerability (CVE-2021-38003) that had been patched in Chrome but remained unaddressed in Sogou's embedded browser, allowing remote code execution with user privileges. This incident highlights the growing sophistication of supply chain attacks targeting widely-deployed software components, particularly those serving large user bases in critical regions. The exploitation of years-old vulnerabilities in embedded browsers demonstrates how legacy code in trusted applications creates persistent attack surfaces for nation-state actors.