Find what you’re looking for

Advanced Android Banking Malware: ToxicPanda 2.0 and GoldDigger Threaten Global Financial Security

ToxicPanda 2.0 represents a significant evolution in Android banking malware, expanding from targeting 16 banking applications to 349 financial institutions across 16 countries. The malware leverages Android accessibility services to steal UI elements, deploy overlay-based credential theft, and abuse Android Wireless Debugging for privilege escalation. Concurrently, GoldDigger banking trojan has launched massive infection campaigns in South Africa and the U.K., impersonating airline companies and retailers while performing sophisticated on-device fraud through real-time screen access and automated transaction manipulation. These incidents highlight the rapidly evolving landscape of mobile banking threats, where attackers are leveraging cloud infrastructure for distribution and implementing advanced evasion techniques. The shift toward on-device fraud capabilities and expanded targeting scope reflects the growing sophistication of mobile threat actors and their ability to adapt to modern security measures.

NASA Spacecraft Control Vulnerability Highlights Critical Infrastructure Security Gaps

In August 2026, Cycode security researchers disclosed critical vulnerabilities in NASA's AIT-GUI spacecraft control software that allowed unauthenticated attackers to issue arbitrary commands to spacecraft and instruments. The flaw chain, rated 9.4 CVSS, affected AIT-GUI versions 2.5.1 and earlier, exposing command endpoints without authentication, authorization, or CSRF protection. Attackers could execute server-side scripts, run command sequences, and issue spacecraft commands via simple HTTP POST requests to the web interface that bound to all network interfaces by default. This incident highlights the growing risk of AI-assisted vulnerability research and the critical need for secure-by-default configurations in operational technology environments. As space infrastructure becomes increasingly digitized and interconnected, authentication gaps in command and control systems represent existential risks to mission-critical operations.

Manic Malware Breaks the Air Gap: How Wi-Fi Mesh Networks Enable Data Theft from Offline Devices

The Manic Android malware campaign emerged in February 2026, targeting Ukrainian banks, government services, and Russian financial institutions through sophisticated phishing sites and dropper applications. This hybrid banking malware and spyware employs a novel Wi-Fi mesh technique that enables infected devices to relay stolen data through nearby compromised devices with internet access, allowing data exfiltration even when the primary device is offline. The malware monitors 169 package IDs across financial, government, and messaging applications, utilizing accessibility services abuse and transparent overlays to capture sensitive data including PIN codes, authentication credentials, and location information. This incident represents a significant evolution in mobile threats, demonstrating how attackers are adapting to air-gapped security measures and developing mesh-based exfiltration techniques. The campaign's timing amid ongoing geopolitical tensions and its focus on Ukrainian infrastructure highlights the intersection of cybercrime and nation-state activities, making mobile device security and network segmentation increasingly critical for organizational defense strategies.

CDN Tsunami Attack Exploits HTTP/3 Protocol Translation for 350x DoS Amplification

In August 2026, cybersecurity researchers disclosed the CDN Tsunami attack, exploiting HTTP/3 to HTTP/1.1 protocol translation vulnerabilities in major CDNs including Cloudflare, Amazon CloudFront, Fastly, Alibaba, Baidu, and Tencent. The attack leverages QPACK header compression and HTTP/3 multiplexing to achieve up to 350x bandwidth amplification against origin servers, requiring minimal attacker resources while consuming over 100 Mbps at the target. The vulnerability affects over 42,000 potentially vulnerable domains and demonstrates how protocol mismatches in CDN architectures create dangerous amplification vectors. This incident highlights the emerging threat landscape around modern web protocols and infrastructure complexity, as organizations increasingly rely on CDNs for performance and protection while inadvertently introducing new attack vectors through protocol translation gaps.

Zombie Card Attack Exposes Critical Flaw in Visa Contactless Payment Security

Researchers at the University of Massachusetts Amherst demonstrated a critical vulnerability in Visa contactless payment systems that allows attackers to revive expired credit cards for fraudulent transactions. The 'Zombie Card' attack exploits a cryptographic binding weakness in Visa's Kernel 3 EMV implementation, enabling attackers with physical access to expired cards and NFC relay equipment to modify expiration dates without breaking card cryptography. Testing across five major US banks showed one bank approved fraudulent transactions up to $500, while others either declined or failed the modification. The attack requires the original account to remain open and relies on issuers not independently verifying expiration dates during authorization, exposing fundamental flaws in contactless payment security architecture. This vulnerability highlights the growing sophistication of payment card fraud techniques as contactless transactions become mainstream, with researchers identifying similar NFC relay malware like WindRelay actively targeting victims across Europe, demonstrating that theoretical academic research quickly translates into real-world criminal exploitation.

SilkParasite APT's Advanced RATs Target Central Asian Governments

In late 2025, the Chinese-nexus advanced persistent threat (APT) group known as SilkParasite initiated a cyber-espionage campaign targeting government organizations across Central Asia, including Uzbekistan, Turkmenistan, Kyrgyzstan, Tajikistan, and Kazakhstan. Utilizing spear-phishing emails with regionally tailored Office documents, often within password-protected RAR archives, the attackers deployed a suite of seven remote access Trojans (RATs), five of which were previously undocumented. These RATs enabled long-term access to sensitive governmental systems, facilitating intelligence gathering and potential disruption of critical operations. This incident underscores the evolving sophistication of state-sponsored cyber threats, particularly the use of modular and AI-assisted malware designed to evade detection. The strategic focus on Central Asian governments highlights a shift in geopolitical cyber-espionage activities, emphasizing the need for enhanced cybersecurity measures and international cooperation to mitigate such threats.

China-Linked AI Cyberattack Targets Taiwan Government in 2026

In early July 2026, a sophisticated cyberattack targeted Taiwan's government agencies and critical infrastructure. Over four days, attackers employed autonomous AI agents to compromise 85 government accounts, exfiltrate over 2,500 personnel records, and infiltrate the nuclear safety agency and multiple energy companies. The AI-driven system utilized open-source frameworks like Hermes and OpenClaw to autonomously map networks, identify vulnerabilities, and adapt strategies in real-time, all while masquerading as legitimate penetration tests. The attack did not rely on zero-day exploits but exploited existing security weaknesses such as exposed APIs and weak authentication mechanisms. Internal communications in Simplified Chinese suggest a high probability of Chinese state-sponsored involvement. This incident underscores the escalating threat of AI-driven cyberattacks, highlighting the need for enhanced identity management and advanced behavioral monitoring to counteract machine-driven intrusions with human-like coordination and minimal oversight.

Unveiling 'CoSnitch': The AI Vulnerability in Microsoft Copilot

In December 2025, Varonis Threat Labs identified a vulnerability in Microsoft Copilot Personal, termed 'CoSnitch,' which allowed attackers to manipulate the AI into revealing its own architectural details. By crafting specific prompts, researchers induced Copilot to disclose information that facilitated memory poisoning, automatic prompt execution via specially crafted URLs, and data exfiltration. Microsoft addressed this issue by releasing patches on August 18, 2026, and confirmed that enterprise customers were unaffected. This incident underscores the evolving threat landscape where AI systems can be exploited to divulge sensitive information. It highlights the necessity for continuous security assessments and the implementation of robust guardrails to prevent similar vulnerabilities in AI-driven platforms.

Critical GitLab Vulnerability CVE-2026-19478: Immediate Action Required

In August 2026, GitLab disclosed a critical vulnerability (CVE-2026-19478) in its Community and Enterprise Editions, affecting versions from 18.2 up to 19.2.3. This code injection flaw within the GraphQL API allows unauthenticated attackers to remotely modify or delete public projects and user data. The vulnerability has been assigned a CVSS score of 9.4 due to its high impact on data integrity and availability. Organizations using self-managed GitLab instances are urged to upgrade to the patched versions 18.11.11, 19.0.8, 19.1.6, or 19.2.4 immediately to mitigate this risk. The disclosure of CVE-2026-19478 underscores the critical importance of securing APIs against unauthorized access and code injection attacks. As threat actors increasingly exploit such vulnerabilities, organizations must prioritize timely patching and implement robust monitoring of API activities to detect and prevent unauthorized operations.

Clop Ransomware's Targeted Attack on PTC Windchill via CVE-2026-12569

In August 2026, the Clop ransomware group exploited a critical vulnerability (CVE-2026-12569) in PTC Windchill and FlexPLM servers, deploying a custom JavaServer Pages (JSP) web shell. This sophisticated implant enabled attackers to decrypt stored credentials, map sensitive engineering data, and execute arbitrary code, facilitating remote access, data exfiltration, and potential ransomware deployment. The attack underscores the evolving tactics of threat actors in targeting enterprise Product Lifecycle Management (PLM) software to access proprietary information and credentials. Organizations utilizing such platforms must prioritize timely patching and robust security measures to mitigate the risk of similar exploits.

Critical Vulnerabilities in macOS, SharePoint, vCenter, and IKE Under Active Exploitation

In August 2026, the U.S. Cybersecurity and Infrastructure Security Agency (CISA) added four critical vulnerabilities to its Known Exploited Vulnerabilities (KEV) catalog, indicating active exploitation in the wild. These vulnerabilities include CVE-2026-65400 affecting Apple macOS, CVE-2026-55040 in Microsoft SharePoint, CVE-2026-59310 in VMware vCenter, and CVE-2026-33824 in Microsoft Internet Key Exchange (IKE) Service Extensions. Exploitation of these flaws has led to unauthorized access, deployment of cryptocurrency miners, backdoors, and ransomware attacks across multiple countries. The active exploitation of these vulnerabilities underscores the persistent threat posed by sophisticated cyber actors targeting widely used enterprise systems. Organizations are urged to prioritize patching and implement robust security measures to mitigate potential risks associated with these exploits.

Microsoft Uncovers Extensive MacSync Stealer Infrastructure

In August 2026, Microsoft Defender Experts identified over 30 web domains associated with MacSync Stealer, a macOS-targeted information-stealing malware. The investigation revealed that the malware utilized social engineering tactics, such as ClickFix, to trick users into executing malicious commands in the Terminal. Once executed, MacSync Stealer collected sensitive data, including macOS Keychain contents, browser credentials, SSH keys, and AWS credentials, which were then exfiltrated to attacker-controlled servers. The malware employed various evasion techniques, including in-memory execution and the use of native macOS utilities, to minimize detection. This incident underscores the evolving sophistication of macOS-targeted malware and the increasing use of social engineering techniques to bypass traditional security measures. Organizations must remain vigilant and educate users about the risks of executing unverified commands, especially as threat actors continue to adapt their methods to exploit human factors.