This article analyzes real-world instances of indirect prompt injection (IDPI) attacks targeting AI agents and large language models integrated into web systems. The researchers identify 22 distinct techniques used by attackers to embed malicious prompts in webpages, including visual concealment, obfuscation, and dynamic execution methods. They categorize attacker intents ranging from low-severity disruptions to critical data destruction attempts. Notable findings include the first observed case of AI-based ad review evasion and attempts at search engine optimization manipulation. The article presents a taxonomy of web-based IDPI attacks and provides insights into attack trends based on telemetry data. The researchers emphasize the need for proactive, web-scale defenses to detect IDPI and distinguish between benign and malicious prompts.
A newly announced quantum algorithm suggests that the practical decryption of RSA encryption using quantum computers could occur much sooner than previously anticipated. Traditionally, the threat to RSA and ECC encryption from quantum computing was linked to Shor’s algorithm, which requires million-qubit quantum computers, a technology not expected in the near term. This new development challenges that assumption, indicating that the quantum breaking point for widely used public key cryptography may be closer. Although no known exploits exist in the wild yet, this advancement raises significant concerns for the security of current cryptographic systems. Organizations relying on RSA and ECC for confidentiality and integrity should begin preparing for a post-quantum cryptographic transition. The threat impacts global communications, financial transactions, and data protection mechanisms. Immediate mitigation involves accelerating the adoption of quantum-resistant algorithms and increasing cryptographic agility. Countries with advanced technological infrastructure and heavy reliance on RSA/ECC encryption are at higher risk. The overall severity is assessed as high due to the potential widespread impact and the fundamental nature of the cryptographic break, despite the current lack of active exploitation.
Researchers have uncovered a Wi-Fi vulnerability that allows nearby attackers to intercept sensitive data and execute machine-in-the-middle attacks against connected devices. The post New ‘AirSnitch’ Attack Shows Wi-Fi Client Isolation Could Be a False Sense of Security appeared first on SecurityWeek .
Hackers stole names, Social Security numbers, driver’s license information, voter registration records, and health-related information. The post 1.2 Million Affected by University of Hawaii Cancer Center Data Breach appeared first on SecurityWeek .
Microsoft has discovered phishing campaigns exploiting OAuth's redirection mechanisms to bypass conventional defenses. Attackers create malicious applications with redirect URIs pointing to malicious domains, then distribute phishing links prompting targets to authenticate. The attack abuses OAuth's error handling to redirect users from trusted providers to attacker-controlled sites for phishing or malware delivery. Campaigns targeted government and public sectors using e-signature, financial, and political lures. Some attacks led to malware downloads and endpoint compromise via PowerShell and DLL side-loading. Mitigation involves governing OAuth apps, limiting user consent, reviewing permissions, and implementing cross-domain detection across email, identity, and endpoint.
A new malicious campaign involving seventeen npm packages has been identified, utilizing Pastebin and text steganography as a dead-drop resolver. The attackers employ a complex decoding mechanism to extract C2 URLs from seemingly benign text on Pastebin. The malware targets multiple platforms, including Windows, macOS, and Linux, downloading and executing platform-specific payloads. The infection chain involves multiple fallback domains hosted on Vercel, demonstrating a sophisticated approach to maintain persistence. This novel technique, along with other recent developments, indicates an accelerated pace of testing and development by the threat actor, suggesting continued iterations in their infection methodologies.
A recent PlugX campaign utilized phishing emails with a 'Meeting Invitation' lure to deploy malware through DLL side-loading. The infection chain begins with a zip file containing a malicious .csproj file and MSBuild executable. The .csproj file downloads three components: a legitimate G DATA Antivirus executable, a malicious Avk.dll (PlugX variant), and an encrypted AVKTray.dat file. The malware uses DLL side-loading, API hashing, and XOR encryption for obfuscation. It establishes persistence via the Run registry key and communicates with a command and control server. The campaign showcases PlugX's continued evolution while maintaining its core characteristics, highlighting its ongoing relevance in cyber-espionage operations.
It's Friday, let's have a look at another simple piece of malware to close a busy week! I received a Fedex notification about a delivery. Usually, such emails are simple phishing attacks that redirect you to a fake login page to collect your credentials. Here, it was a bit different:
Threat actors have discovered a novel method to bypass security controls by abusing the .arpa top-level domain (TLD) in conjunction with IPv6 tunnels. They are exploiting a feature in DNS record management of certain providers to add IP address records for .arpa domains, allowing them to host phishing content on domains that should not resolve to an IP address. The phishing campaigns use spam emails impersonating major brands, with hyperlinked images leading to malicious websites through traffic distribution systems. This technique weaponizes trusted infrastructure essential for network operations, making it challenging for security tools to detect suspicious domains based on reputation, registration information, or policy blocklists.
This threat report highlights four critical risk areas that organizational boards must actively manage rather than ignore as background noise. The focus is on resilience—ensuring business continuity even when cyberattacks succeed—rather than attempting to prevent every possible attack. Although no specific vulnerability details, affected versions, or exploits are provided, the medium severity indicates notable concern. The absence of concrete technical indicators or patches suggests this is a strategic risk advisory rather than a discrete technical vulnerability. Organizations should prioritize risk management frameworks that emphasize detection, response, and recovery capabilities. This approach aligns with modern cybersecurity paradigms that accept breach inevitability and focus on minimizing operational impact. The threat is relevant globally, especially for enterprises with significant digital assets and complex supply chains. Given the lack of direct exploitability and technical specifics, the suggested severity is medium, reflecting moderate risk to confidentiality, integrity, and availability without immediate exploitation evidence.
The UNC2814 threat actor, active since at least 2017, has been conducting a Chinese cyberespionage campaign targeting telecommunications companies and government organizations across 42 countries. Google recently disrupted this campaign, which focused on espionage activities rather than destructive attacks. The threat actor's operations involve sophisticated techniques to infiltrate and maintain persistence within targeted networks. Although no specific vulnerabilities or exploits have been publicly disclosed, the campaign's longevity and broad geographic scope highlight its strategic nature. The medium severity rating reflects the espionage focus, limited public exploit details, and absence of widespread destructive impact. Organizations in telecommunications and government sectors worldwide should remain vigilant against such advanced persistent threats. Mitigation requires enhanced threat detection, network segmentation, and intelligence sharing. Countries with significant telecom infrastructure and geopolitical interest in China are at higher risk. Overall, this campaign underscores the ongoing cyberespionage risks posed by state-sponsored actors targeting critical sectors globally.
In January 2026, a new variant of the PlugX malware was observed being used in targeted attacks. Analysis suggests involvement of the UNC6384 APT group, linked to Mustang Panda, targeting government agencies in Southeast Asia. The malware uses a browser updater disguise to download and execute a malicious MSI file, leading to PlugX infection. The STATICPLUGIN downloader uses a revoked code-signing certificate from a Chinese company. The PlugX variant employs DLL sideloading and shellcode execution techniques. Its configuration is encrypted using RC4 and custom encoding. C2 servers were identified as fruitbrat[.]com and 108.165.255[.]97:443. The ongoing improvements to PlugX indicate its continued use in targeted attacks by APT groups.
Wynn Resorts, a prominent casino and hotel operator, has confirmed a data breach involving the theft of employee data by the hacking group ShinyHunters. The attackers initially posted the stolen data on a leak site but later removed it, indicating possible negotiations or other undisclosed reasons. Although no known exploits are currently active in the wild, the breach exposes sensitive employee information, raising concerns about privacy and potential misuse. The incident highlights vulnerabilities in Wynn Resorts' data security and the ongoing threat posed by cybercriminal groups targeting hospitality and entertainment sectors. Organizations with similar profiles should be vigilant and review their security postures to prevent similar breaches.
