Search This Blog

Powered by Blogger.

Blog Archive

Labels

Footer About

Footer About

Labels

Threat Actors Exploit Fortinet Devices and Steal Firewall Configurations


Fortinet products targeted

Threat actors are targeting Fortinet FortiGate devices via automated attacks that make rogue accounts and steal firewall settings info. 

The campaign began earlier this year when threat actors exploited an unknown bug in the devices’ single-sign-on (SSO) option to make accounts with VPN access and steal firewall configurations. This means automation was involved. 

About the attack

Cybersecurity company Arctic Wolf discovered this attack and said they are quite similar to the attacks it found in December after the reveal of a critical login bypass flaw (CVE-2025-59718) in Fortinet products. 

The advisory comes after a series of reports from Fortinet users about threat actors abusing a patch bypass for the bug CVE-2025-59718 to take over patched walls. 

Impacted admins complaint that Fortinet said that the latest FortiOS variant 7.4.10 doesn't totally fix the authentication bypass bug, which should have been fixed in December 2025.

Patches and fixing 

Fortinet also plans on releasing more FortiOS variants soon to fully patch the CVE-2025-59718 security bug. 

Following an SSO login from cloud-init@mail.io on IP address 104.28.244.114, the attackers created admin users, according to logs shared by impacted Fortinet customers. This matches indications of compromise found by Arctic Wolf during its analysis of ongoing FortiGate attacks and prior exploitation the cybersecurity firm noticed in December. 

Turn off FortiCloud SSO to prevent intrusions. 

Turning off SSO

Admins can temporarily disable the vulnerable FortiCloud login capability (if enabled) by navigating to System -> Settings and changing "Allow administrative login using FortiCloud SSO" to Off. This will help administrators safeguard their firewalls until Fortinet properly updates FortiOS against these persistent assaults.

You can also run these commands from the interface:

"config system global

set admin-forticloud-sso-login disable

end"

What to do next?

Internet security watchdog Shadowserver is investigating around 11,000 Fortinet devices that are vulnerable to online threats and have FortiCloud SSO turned on. 

Additionally, CISA ordered federal agencies to patch CVE-2025-59718 within a week after adding it to its list of vulnerabilities that were exploited in attacks on December 16.

ACF Plugin Flaw Exposes 50,000 WordPress Sites to Admin Takeover

 

A critical vulnerability in the Advanced Custom Fields: Extended (ACF Extended) WordPress plugin has exposed around 50,000 sites to potential hacker takeovers. Tracked as CVE-2025-14533, this flaw affects versions up to 0.9.2.1 and allows unauthenticated attackers to gain administrator privileges through flawed user creation forms. Discovered by researcher Andrea Bocchetti and reported via Wordfence on December 10, 2025, the issue was swiftly patched in version 0.9.2.2 just four days later. Despite the quick fix, download stats show many sites remain unpatched, leaving them vulnerable to remote exploitation.

The vulnerability originates in the plugin's 'Insert User / Update User' form action, where role restrictions are not properly enforced. Attackers can exploit this by submitting crafted requests that assign the 'administrator' role, bypassing any configured limitations in field settings.This privilege escalation requires sites to use forms with a 'role' field mapped to custom fields, a common setup for user registration features. Once successful, hackers achieve full site control, enabling data theft, malware injection, or backdoor installation without needing prior access.

ACF Extended, active on over 100,000 WordPress installations, builds on the popular Advanced Custom Fields plugin to offer developers advanced customization tools. Its widespread use amplifies the risk, as roughly half of users have yet to update since the patch release in mid-December 2025. WordPress sites relying on these plugins for dynamic content often overlook such configurations, inadvertently creating attack vectors.

This privilege escalation bug allows attackers to arbitrarily assign the 'administrator' role during user registration or updates, bypassing any configured limitations in field settings. Exploitation requires sites using ACF Extended forms with a 'role' field mapped to custom fields, a common setup for advanced user management in custom themes and plugins. Once exploited, hackers gain full control, enabling them to install malicious code, steal data, or pivot to server-level compromises without needing credentials.

Threat intelligence from GreyNoise reveals aggressive reconnaissance scanning 706 WordPress plugins, including ACF Extended, by nearly 1,000 IPs across 145 ASNs from late October 2025 to mid-January 2026. While no confirmed exploits of CVE-2025-14533 have surfaced, patterns mirror attacks on vulnerabilities like those in Post SMTP and LiteSpeed Cache, signaling imminent danger.This enumeration boom underscores how attackers probe for unpatched flaws before launching mass campaigns.

Site owners must urgently update to ACF Extended 0.9.2.2 or later via the WordPress dashboard and audit forms for role mappings.Additional steps include disabling public registration, reviewing user accounts for anomalies, and deploying firewalls like Wordfence for real-time blocking. In WordPress's vast ecosystem, proactive patching remains the frontline defense against such admin takeovers, preventing potential site-wide devastation.

Unsecured Database Exposes 149 Million Logins Linked to Infostealer Malware Operations

 

Appearing without warning on the internet, a massive collection of personal login details became reachable to any passerby. This trove - spanning about 96 gigabytes - included close to 150 million distinct credentials gathered from various sources. Not shielded by locks or scrambled coding, its contents lay fully exposed. Inside, endless spreadsheets paired emails with user handles, access codes, plus entry points to accounts. Examination showed evidence of widespread digital theft, driven by aggressive software designed to harvest private information. Such leaks reveal how deeply automated attacks now penetrate everyday online activity. 

Credentials came from people across the globe, tied to many different websites. Access information showed up for big social networks, romance apps, subscription video sites, games, and money-handling services. Among them: login pairs for digital currency storage, bank entry points, and systems linked to payment cards. A mix like that points not to one hacked business but likely stems from software designed to gather passwords automatically.  

What stood out most was the appearance of login details tied to government-backed email addresses in various nations. Though these accounts do not always grant entry to critical infrastructure, basic official credentials might still be exploited - serving as tools for focused scams or fake identities. Starting from minor access points, attackers could work their way deeper into secure environments. The level of danger shifts with each individual's privileges; when higher-access .gov logins fall into the wrong hands, consequences can stretch well beyond a single agency. 

Appearing first in the analysis was a database organized much like those seen in infostealer activities. Keylog results sat alongside extra details - hostnames flipped intentionally to sort thefts by target and origin. Though built on hashes, every record carried its own distinct ID, likely meant to prevent repeats while easing bulk sorting tasks. From this setup emerges something functional: a system shaped for gathering, handling, even passing along login information. Last noted - the traits match what supports credential trafficking behind the scenes. 

With unclear responsibility for the database, reporting went straight to the hosting company. Still, fixing the issue dragged on - weeks passed, with multiple alerts needed before entry was blocked. While delays continued, more data kept flowing in, expanding the volume of sensitive records exposed. Who controlled the system, how long it stayed open online, or whether others harvested its contents stays unanswered. One wrong move here leads to serious trouble. 

When hackers get full logins alongside active URLs, they run automated break-ins across many accounts - this raises chances of stolen identities, fake messages that seem real, repeated fraud, and unauthorized access. Personal habits emerge through used platforms, painting a clearer picture of who someone is online, which deepens threats to private data and future safety. 

Midway through this event lies proof: stealing login details now operates like mass production, fueled by weak cloud setups. Because information-harvesting software grows sharper every month, staying protected means doing basics well - shielding devices, practicing careful habits online, using separate codes everywhere, while adding extra identity checks. Found gaps here reveal something odd at first glance - not just legitimate systems fail from poor setup, but illegal networks do too; when they collapse, masses of people get caught unaware, their private pieces scattered without knowing a breach ever happened.

China-Linked DKnife Threat Underscores Risks to Network Edge Devices

 


Despite adversaries increasing their focus on the network edge, recent findings suggest a sustained and deliberate effort to weaponize routing infrastructure itself for surveillance and delivery purposes. An attacker can observe, modify, and selectively redirect data streams in transit by embedding malicious logic directly into traffic paths rather than relying on endpoint compromise. 

This evolution is reflected in the development of the DKnife framework, which has transformed attacker-in-the-middle capabilities into modular, long-lived platforms that are designed to be persistent, stealthy, and operationally flexible. 

Through the framework's ability to operate at a level where legitimate traffic aggregation and inspection already take place, the line between benign network functionality and hostile control is blurred, enabling malware deployment and long-term monitoring across a variety of device classes and user environments targeted at targeted users. 

According to cybersecurity researchers, DKnife is an adversary-in-the-middle framework that has operated from at least 2019 to maintain router-centric infrastructure by threat actors who have been found to be linked to China. 

In order to enable deep packet inspection, selective traffic manipulation, and covert delivery of malicious payloads, seven Linux-based implants are installed on gateways and edge devices. Several code artifacts and telemetry indicate a clear focus on Chinese-speaking users, including credential-harvesting components tailored specifically for Chinese email services, data exfiltration modules specifically targeted at popular mobile applications, and hard-coded references to domestic media domains buried within the implants. 

It is argued that DKnife's potential strategic value lies in its ability to act as a conduit between legitimate update and download channels and users. As the framework intercepts binary transfers and mobile application updates in transit, it is possible to deploy and manage established backdoors across a broad range of endpoints ranging from desktop systems to mobile devices to Internet of Things environments, including ShadowPad and DarkNimbus. 

According to Cisco Talos, the activity has been associated with the ongoing tracking of a Chinese threat cluster dubbed Earth Minotaur, previously associated with exploit kits like MOONSHINE as well as backdoors like DarkNimbus. The reuse of DarkNimbus is noteworthy, as the malware has also been found in operations attributed to another Chinese advanced persistent threat group, The Wizards, indicating the possibility of sharing tools or infrastructure among these groups. 

Upon further analysis of the infrastructure, it was revealed that DKnife-associated resources overlapped with those connected to WizardNet, a Windows implant deployed by TheWizards through an AitM framework called Spellbinder, which was publicized in 2025. This led to additional connections between DKnife-associated systems and WizardNet resources. 

As Cisco cautions, current insights into DKnife's targeting may be incomplete due to the fact that the configuration data obtained from a single command-and-control server provide limited information about its target market of Chinese-speaking users. It is possible that parallel servers exist to support operations in other regions as well. 

Due to The Wizards' history of targeting individuals and gambling-related entities across Southeast Asia, Greater China, and the Middle East, the convergence of infrastructure and tactics is significant, highlighting the wider implications of DKnife as a traffic hijacking platform with reusable, regionally adaptable features. 

Although researchers have not determined the exact vector used to compromise network equipment, researchers have established that DKnife functions to deliver and control backdoors known as ShadowPad and DarkNimbus, both of which have been used by Chinese-allied threat actors for decades. A technical analysis reveals that there are seven discrete modules in the framework. 

Each module is designed to support a particular operational role, such as traffic inspection, manipulation, and control-and-control messages, as well as origin obfuscation. In addition to packet inspection and attack logic, the system includes relay services to facilitate communication with remote C2 servers as well as a customized reverse proxy derived from HAProxy to mask and manage malicious traffic flows. 

Additionally, DKnife extends its capabilities beyond passive monitoring with additional modules. An attacker is able to establish a virtual Ethernet TAP interface on the compromised router and connect it directly to the local network, effectively placing themselves in the data path of internal communications.

In addition, there are third parties who provide peer-to-peer VPN connectivity using modified n2n software, coordinate the download and update of malicious Android applications, and manage the deployment of the DKnife implants themselves. 

Together, these elements provide a range of tools for a wide range of activities, including DNS hijacking, intercepting legitimate binary and application updates, selectively disrupting security-related traffic, and exfiltrating detailed user activity to external command infrastructures. In addition to intercepting and rewriting packets destined for their original hosts once activated on a device, DKnife also uses its network-bridging capabilities to substitute malicious payloads during transit transparently. 

Through this technique, weaponized APK files can be delivered to Android devices as well as compromised binaries to Windows systems connected to the affected network using this technique. Research conducted by Cisco Talos demonstrated instances in which the framework first installed ShadowPad backdoors for Windows, signed by Chinese certificates, followed by the installation of DarkNimbus backdoors to establish long-term access. 

Unlike secondary droppers, DarkNimbus was delivered directly to Android environments through the manipulated update channel. It was further revealed by investigators that infrastructure was associated with a framework hosting the WizardNet backdoor, a Windows implant previously associated with Spellbinder AitM. This confirmed the link between DKnife and previously documented adversary-in-the-middle attacks. 

Incorporating these tools within the same operational environment implies that development resources will likely be shared or infrastructure will be coordinated. As a result, threat actors are becoming increasingly sophisticated in their use of compromised network devices as covert malware distribution channels as opposed to utilizing endpoints to spread malware. 

The Cisco Talos team further concluded that DKnife is capable of intercepting Windows binary downloads in addition to mobile ecosystems. As observed, the framework was capable of manipulating download URLs in transit, either substituting legitimate installers for trojanized counterparts or redirecting users to malicious distribution points controlled by the attackers. 

In combination with its DNS manipulation capabilities and control over application update channels, DKnife provides an extensive traffic-hijacking platform that can silently deliver malware while maintaining the appearance of normal network behavior.

The framework's components work together to create a continuous attack system at the network gateway that functions in conjunction with each other. Moreover, DKnife offers a broad range of secondary functionality in addition to payload delivery, such as credential harvesting through decrypted POP3 and IMAP sessions, hosting phishing pages, selectively disrupting antivirus and security product traffic, and detailed user activity monitoring. 

Several applications and services were observed to collect telemetry, including messaging platforms, navigation tools, news consumption, telephony, ridesharing, and online shopping, by researchers. In particular, WeChat was observed to receive significant attention, with the framework tracking voice and video calls, message content, media exchanges, and articles accessed through the application. The placement of DKnife on gateway devices permits near real-time visibility into user behavior. 

Activity events are processed internally across the framework's modular components first before being exfiltrated via structured HTTP POST requests to dedicated API endpoints and then forwarded to remote command-and-control infrastructure. 

A significant reduction in the need for persistent malware on individual endpoints is achieved through this architecture, which allows attackers to correlate traffic flows and user actions as packets traverse the network. Researchers note that this approach reflects a greater trend towards infrastructure-level compromise, which is the use of routers and edge devices as persistent delivery platforms for malware. 

According to Cisco Talos, DKnife-associated command-and-control servers remain active as of January 2026, highlighting the continued nature of this threat. An exhaustive set of indicators of compromise has been developed by the firm to assist defenders in identifying compromised systems, as well as emphasizing the need to pay increased attention to network infrastructure as adversaries continue to utilize its unique position within modern digital environments to their advantage.

Spain’s Science Ministry Partially Shuts Online Systems After Suspected Cyber Incident

 



Spain’s Ministry of Science, Innovation and Universities has temporarily disabled parts of its digital infrastructure following what it described as a technical problem. The disruption has affected several online services used by citizens, universities, researchers, and businesses for official procedures and submissions. These platforms support important administrative functions and process sensitive information, which is why access was restricted as a precaution.

The ministry oversees national science policy, research programs, innovation initiatives, and higher education administration. Its systems handle high-value data, including academic and research records, application materials, and personal information linked to students and professionals. Because of the incident, multiple digital services were made unavailable, and active procedures were placed on hold to limit any potential risk to data or system integrity.

In a public notice on its official website, the ministry stated that the incident is under technical assessment and did not disclose further details at the time. The announcement clarified that the ministry’s online portal is only partially operational and that ongoing administrative processes have been paused to protect the rights and lawful interests of affected users. To reduce the impact of the outage, authorities confirmed that deadlines for affected procedures will be extended in line with Spain’s administrative law provisions, so applicants and institutions are not penalized for delays caused by the shutdown.

Separately, claims surfaced on underground online platforms from an individual alleging unauthorized access to the ministry’s systems. The person shared what they presented as sample data to support the claim and stated that additional information was available for sale. The material reportedly includes personal records, email information, application-related documents, and images of official paperwork. These claims have not been independently verified, and the online space where the samples were shared later became inaccessible.

The same individual alleged that access was gained by exploiting a security weakness that can allow users to reach restricted resources without proper authorization. Such flaws, when present in web applications, can expose internal systems if not properly secured. At this stage, the technical details of the claim remain unconfirmed by authorities.

Spanish media outlets have reported that a ministry spokesperson acknowledged that the service disruption is linked to a cybersecurity incident. However, officials have not confirmed whether any data was accessed or taken, nor have they outlined the scope of any potential compromise. The ministry has indicated that investigations are ongoing to determine what occurred and to restore services safely.

Cybersecurity experts consistently warn that public sector systems are frequent targets because of the volume and sensitivity of data they manage. Strong access controls, continuous monitoring, and timely security updates are critical to reducing exposure to such risks. Further updates from the ministry are expected once technical assessments are completed and the situation is fully clarified.

Infy Hackers Strike Again With New C2 Servers After Iran's Internet Shutdown Ends


Infy group's new attack tactic 

An Iranian hacking group known as Infy (aka Prince of Persia) has advanced its attack tactics to hide its operations. The group also made a new C2 infrastructure while there was a wave of internet shutdown imposed earlier this year. The gang stopped configuring its C2 servers on January 8 when experts started monitoring Infy. 

In reaction to previous protests, Iranian authorities implemented a nationwide internet shutdown on this day, which probably indicates that even government-affiliated cyber units did not have the internet. 

About the campaign 

The new activity was spotted on 26 January 2026 while the gang was setting up its new C2 servers, one day prior to the Iranian government’s internet restrictions. This suggests that the threat actor may be state-sponsored and supported by Iran. 

Infy is one of the many state-sponsored hacking gangs working out of Iran infamous for sabotage, spying, and influence campaigns coordinated with Tehran’s strategic goals. However, it also has a reputation for being the oldest and less famous gangs staying under the radar and not getting caught, working secretly since 2004 via “laser-focused” campaigns aimed at people for espionage.

The use of modified versions of Foudre and Tonnerre, the latter of which used a Telegram bot probably for data collection and command issuance, were among the new tradecraft linked to the threat actor that SafeBreach revealed in a report released in December 2025. Tornado is the codename for the most recent version of Tonnerre (version 50).

The report also revealed that threat actors replaced the C2 infrastructure for all variants of Tonnerre and Foudre and also released Tornado variant 51 that employs both Telegram and HTTP for C2.

It generates C2 domain names using two distinct techniques: a new DGA algorithm initially, followed by fixed names utilizing blockchain data de-obfuscation. We believe that this novel method offers more flexibility in C2 domain name registration without requiring an upgrade to the Tornado version.

Experts believe that Infy also abused a 1-day security bug in WinRAR to extract the Tornado payload on an infected host to increase the effectiveness of its attacks. The RAR archives were sent to the Virus Total platform from India and Germany in December 2025. This means the two countries may have been victims. 



AISURU/Kimwolf Botnet Behind Record 31.4 Tbps DDoS Attack, Cloudflare Reveals

 

A massive distributed denial-of-service (DDoS) assault reaching an unprecedented peak of 31.4 terabits per second (Tbps) has been attributed to the AISURU/Kimwolf botnet. The attack, which lasted just 35 seconds, is now being described as one of the largest hyper-volumetric DDoS events ever recorded.

Cloudflare said it automatically identified and blocked the activity, noting that the incident was part of a wider surge in hyper-volumetric HTTP DDoS attacks linked to AISURU/Kimwolf during the fourth quarter of 2025. The specific attack occurred in November 2025.

The botnet has also been associated with a separate campaign dubbed The Night Before Christmas, which began on December 19, 2025. According to Cloudflare, attacks observed during this campaign averaged 3 billion packets per second (Bpps), 4 Tbps, and 54 million requests per second (Mrps). At their peak, the attacks escalated to 9 Bpps, 24 Tbps, and 205 Mrps.

"DDoS attacks surged by 121% in 2025, reaching an average of 5,376 attacks automatically mitigated every hour," Cloudflare's Omer Yoachimik and Jorge Pacheco said. "In 2025, the total number of DDoS attacks more than doubled to an incredible 47.1 million."

The web infrastructure firm reported mitigating 34.4 million network-layer DDoS attacks throughout 2025, a sharp increase from 11.4 million in 2024. In the final quarter of 2025 alone, network-layer incidents represented 78% of all DDoS activity. Overall, DDoS attacks climbed 31% quarter-over-quarter and rose 58% compared to the previous year. 

Hyper-volumetric DDoS attacks also saw a significant rise, increasing by 40% in Q4 2025 compared to the previous quarter, jumping from 1,304 to 1,824 incidents. Earlier in the year, Q1 2025 recorded 717 such attacks. Alongside the growing frequency, the scale of these attacks expanded dramatically, with sizes increasing by more than 700% compared to large-scale incidents observed in late 2024.

AISURU/Kimwolf is believed to have compromised over 2 million Android devices, largely unbranded Android TVs, which were absorbed into its botnet. Many of these infections were facilitated through residential proxy networks such as IPIDEA. In response, Google recently disrupted the proxy service and initiated legal action to dismantle dozens of domains used to manage infected devices and route proxy traffic.

Google also collaborated with Cloudflare to interfere with IPIDEA’s domain resolution capabilities, significantly weakening the operators’ command-and-control infrastructure.

“As part of the Google-led disruption effort, Cloudflare participated by suspending access to many accounts and domains that were misusing its infrastructure," Cloudflare told The Hacker News over email. "Threat actors were attempting to distribute malware and provide markets for people seeking access to the network of illicit residential proxies."

Investigations suggest that IPIDEA recruited infected devices using at least 600 malicious Android applications embedded with proxy SDKs, along with more than 3,000 trojanized Windows executables masquerading as OneDriveSync tools or Windows updates. The Beijing-based firm has also promoted VPN and proxy applications that covertly transformed users’ Android devices into proxy exit nodes without their awareness or permission.

Additionally, threat actors have been identified operating more than a dozen residential proxy services posing as legitimate businesses. These offerings, despite appearing separate, are all reportedly connected to a centralized infrastructure controlled by IPIDEA.

Cloudflare highlighted several additional trends observed during Q4 2025. Telecommunications companies, service providers, and carriers were the most targeted industries, followed by IT services, gambling, gaming, and software sectors. The most attacked countries included China, Hong Kong, Germany, Brazil, the United States, the United Kingdom, Vietnam, Azerbaijan, India, and Singapore.

Bangladesh overtook Indonesia as the largest source of DDoS traffic globally, with Ecuador, Indonesia, Argentina, Hong Kong, Ukraine, Vietnam, Taiwan, Singapore, and Peru also ranking among the top origins of attack traffic.

"DDoS attacks are rapidly growing in sophistication and size, surpassing what was previously imaginable," Cloudflare said. "This evolving threat landscape presents a significant challenge for many organizations to keep pace. Organizations currently relying on on-premise mitigation appliances or on-demand scrubbing centers may benefit from re-evaluating their defense strategy."

A Quiet Breach of a Familiar Tool, Notepad++

For six months last year the update system of Notepad++, one of the world’s most widely used Windows text editors, was quietly subverted by hackers linked by investigators to the Chinese state. The attackers used their access not to disrupt the software openly, but to deliver malicious versions of it to carefully chosen targets. 

According to a statement published this week on the project’s official website, the intrusion began in June with an infrastructure-level compromise that allowed attackers to intercept and redirect update traffic meant for notepad-plus-plus.org. Selected users were silently diverted to rogue update servers and served backdoored versions of the application. Control over the update infrastructure was not fully restored until December. 

The developers said the attackers exploited weaknesses in how older versions of Notepad++ verified updates. By manipulating traffic between users and the update servers, they were able to substitute legitimate downloads with malicious ones. 

Although update packages were signed, earlier design choices meant those signatures were not always robustly checked, creating an opening for tampering by a well-resourced adversary. Security researchers say the campaign was highly targeted. 

The attackers installed a previously unknown backdoor, dubbed Chrysalis, which Rapid7 described as a custom and feature-rich tool designed for persistent access rather than short-term disruption. Such sophistication suggests strategic objectives rather than criminal opportunism. 

Independent researcher Kevin Beaumont reported that several organisations with interests in East Asia experienced hands-on intrusions linked to compromised Notepad++ installations, indicating that attackers were able to take direct control of affected systems. 

He had raised concerns months earlier after a Notepad++ update quietly strengthened its updater against hijacking. The episode underlines a broader vulnerability in the global software supply chain. Open-source tools such as Notepad++ are deeply embedded in corporate and government systems, yet are often maintained with limited resources. That imbalance makes them attractive targets for state-backed hackers seeking discreet access rather than noisy disruption. 

Notepad++ developers have urged users to update manually to the latest version and large organisations to consider restricting automated updates. The incident also serves as a reminder that even modest, familiar software can become a conduit for serious espionage when its infrastructure is neglected.

AI Hijacks AWS Cloud in 8 Minutes via Exposed Keys

 

An AI-assisted cyberattack hijacked a company's AWS cloud infrastructure in just eight minutes after attackers discovered exposed test credentials in a public S3 bucket, demonstrating how configuration errors can fuel lightning-fast breaches in the era of automated threats. This incident, uncovered by Sysdig's Threat Research Team on November 28, 2025, exposed vulnerabilities in cloud access management and the growing role of large language models (LLMs) in offensive operations.

The breach began with a simple oversight: credentials named with "AI" references sat openly in an S3 bucket, ripe for discovery during routine scans. Despite a ReadOnlyAccess policy limiting initial access, the intruder launched a massive enumeration campaign, probing Secrets Manager, RDS databases, and CloudWatch logs to blueprint the entire environment without raising alarms. This reconnaissance phase set the stage for rapid escalation, underscoring how even restricted keys can serve as footholds for deeper intrusions.

Attackers then pivoted to code injection on Lambda functions, iteratively tampering with one called EC2-init until they commandeered an account named "frick," granting full administrative privileges. They compromised 19 distinct AWS principals, enabling abuse of Bedrock AI models like Claude 3.5 Sonnet and DeepSeek R1, alongside attempts to launch a "stevan-gpu-monster" GPU instance that could have racked up £18,000 ($23,600) in monthly costs. Sysdig researchers identified LLM hallmarks, including Serbian-commented code, hallucinated AWS IDs like "123456789012," and phantom GitHub references, confirming AI's hand in accelerating the assault.

To evade detection, the threat actor cycled through an IP rotator and 19 identities, attempting lateral movement via default roles like OrganizationAccountAccessRole in a multi-account setup. This stealthy persistence highlights evolving tactics where AI not only speeds execution but also enhances obfuscation, turning minutes-long attacks into prolonged threats if undetected.

Experts warn that mundane errors like exposed keys—not novel exploits—drive such incidents, urging organizations to ditch static credentials for short-lived IAM roles, harden automated accounts, and monitor for anomalous enumeration spikes. As breaches shrink from days to minutes, AI-aware defenses must match this pace to protect cloud assets effectively.

Threat Actors Leverage Hugging Face to Spread Android Malware at Scale


 

Initially appearing as a routine security warning for mobile devices, this warning has evolved into a carefully engineered malware distribution pipeline. Researchers at Bitdefender have identified an Android campaign utilizing counterfeit security applications that serve as the first stage droppers for remote access Trojans, known as TrustBastion. 

The operators have opted not to rely on traditional malware hosting infrastructure, but have incorporated their delivery mechanism into Hugging Face's public platform, allowing it to conceal malicious activity through its reputation and traffic profile. 

Social engineering is used to drive the infection chain, with deceptive ads and fabricated threat alerts causing users to install the malware. The app silently retrieves a secondary payload from Hugging Face once it has been installed on the device, providing persistence via extensive permission abuse. 

At scale, the campaign is distinguished by a high degree of automation, resulting in thousands of distinct Android package variants, thereby evading signature-based detection and complicating attribution, thus demonstrating the shift toward a more industrialized approach to mobile malware. 

Using this initial foothold as a starting point, the campaign illustrates how trusted developer infrastructure can be repurposed to support a large-scale theft of mobile credentials. As a consequence, threat actors have been using Hugging Face as a distribution channel for thousands of distinct Android application packages that were designed to obtain credentials related to widely used financial, banking, and digital payment services.

Generally, Hugging Face is regarded as a low-risk domain, meaning that automated security controls and suspicion from users are less likely to be triggered by this site's hosting and distribution of artificial intelligence, natural language processing, and machine learning models.

Despite the fact that the platform has previously been abused to host malicious AI artifacts, Bitdefender researchers point out that its exploitation as a delivery channel for Android malware constitutes an intentional attempt to disguise the payload as legitimate development traffic. It has been determined that the infection sequence begins with the installation of an application disguised as a mobile security solution known as TrustBastion. 

Using scareware-style advertisements, the app presents fake warnings claiming that the device has been compromised, urging immediate installation to resolve alleged threats, including phishing attempts, fraudulent text messages, and malware. 

Upon deployment, the application displays a mandatory update prompt which is closely similar to that of Google Play, thereby reinforcing the illusion of legitimacy. In lieu of embedding malicious code directly, the dropper contacts infrastructure associated with the trustbastion[.]com domain, which redirects the user to a repository containing Hugging Face datasets. 

After retrieving the final malicious APK via Hugging Face's content delivery network, the attackers complete a staged payload delivery process that complicates detection and allows them to continuously rotate malware variants with minimal operational overhead, complicating detection. This stage demonstrates why Hugging Face was purposefully integrated into the attacker's delivery chain during this phase of the operation. 

It is common for security controls to flag traffic from newly registered or low-reputation domains quickly, causing threat actors to route malicious activity through well-established platforms that blend into normal network behavior, resulting in the use of well-established platforms.

TrustBastion droppers are not designed to retrieve spyware directly from attacker-controlled infrastructure in this campaign. Rather than hosting the malware itself, it initiates a request to a website associated with the trustbastion[. ]com domain, which serves as an intermediary rather than as a hosting point for it.

The server response does not immediately deliver a malicious application package. The server returns a HTML resource that contains a redirect link to a Hugging Face repository where the actual malware can be found. By separating the initial contact point from the final malware host, the attackers introduce additional indirection, which makes static analysis and takedown efforts more challenging. 

According to Bitdefender, the malicious datasets were removed after being notified by Hugging Face before publication of its findings. Telemetry indicates the campaign had already reached a significant number of victims before the infrastructure was dismantled, despite the swift response. Furthermore, analysis of the repositories revealed unusually high levels of activity over a short period of time. 

A single repository accumulated over 6,000 commits within a month, indicating that it was fully automated. A new payload was generated and committed approximately every 15 minutes, according to Bitdefender. A number of repositories were taken offline during the campaign, but the campaign displayed resilience by reappearing under alternative redirect links, using the same core codebase and only minor cosmetic changes to the icons and application metadata. 

The operators further undermined traditional defense effectiveness by utilizing polymorphic techniques throughout the payloads they used. The uploaded APKs were freshly constructed, retaining identical malicious capabilities while introducing small structural changes intended to defeat hash-based detection. 

It was noted by Bitdefender that this approach increased evasion against signature-driven tools, but that the malware variants maintained consistent behavioral patterns, permission requests, and network communication traits, which made them more susceptible to behavioral and heuristic analysis in the future. 

After installation, the malware presents itself as a benign "Phone Security" feature and guides users through the process of enabling Android Accessibility Services. This step allows the remote access trojan to obtain extensive information about user activity and on-screen activity. In order to monitor activity in real time, capture sensitive screen content, and relay information to the malware's command and control servers, additional permissions are requested. 

By impersonating legitimate financial and payment applications, such as Alipay and WeChat, this malware enhances the threat. By intercepting credentials and collecting lock-screen verification information, it becomes a full-spectrum tool to collect credentials and spy on mobile devices. 

In a defensive perspective, this campaign reminds us that trust in popular platforms can be strategically exploited if security assumptions are not challenged. By combining legitimate developer infrastructure abuse with high levels of automation and polymorphic payload generation, traditional indicators alone cannot detect these types of attacks. 

For Bitdefender's users, the findings reinforce the importance of identifying such threats earlier in the infection chain through behavioral analysis, permission monitoring, and anomaly-based network inspection. Users are advised to take precautions when responding to unsolicited security alerts or applications requesting extensive system privileges based on the findings.

Additionally, the operation highlights the growing adoption of cloud-native distribution models by malicious mobile malware actors, emphasizing the importance of platform providers, security vendors, and enterprises collaborating more closely to monitor abuse patterns and respond quickly to emerging misuses of trusted ecosystems.

Orchid Security Launches Tool to Monitor Identity Behavior Across Business Applications

 



Modern organizations rely on a wide range of software systems to run daily operations. While identity and access management tools were originally designed to control users and directory services, much of today’s identity activity no longer sits inside those centralized platforms. Access decisions increasingly happen inside application code, application programming interfaces, service accounts, and custom login mechanisms. In many environments, credentials are stored within applications, permissions are enforced locally, and usage patterns evolve without formal review.

As a result, substantial portions of identity activity operate beyond the visibility of traditional identity, privileged access, and governance tools. This creates a persistent blind spot for security teams. The unseen portion of identity behavior represents risk that cannot be directly monitored or governed using configuration-based controls alone.

Conventional identity programs depend on predefined policies and system settings. These approaches work for centrally managed user accounts, but they do not adequately address custom-built software, legacy authentication processes, embedded secrets, non-human identities such as service accounts, or access routes that bypass identity providers. When these conditions exist, teams are often forced to reconstruct how access occurred after an incident or during an audit. This reactive process is labor-intensive and does not scale in complex enterprise environments.

Orchid Security positions its platform as a way to close this visibility gap through continuous identity observability across applications. The platform follows a four-part operational model designed to align with how security teams work in practice.

First, the platform identifies applications and examines how identity is implemented within them. Lightweight inspection techniques review authentication methods, authorization logic, and credential usage across both managed and unmanaged systems. This produces an inventory of applications, identity types, access flows, and embedded credentials, establishing a baseline of how identity functions in the environment.

Second, observed identity activity is evaluated in context. By linking identities, applications, and access paths, the platform highlights risks such as shared or hardcoded secrets, unused service accounts, privileged access that exists outside centralized controls, and differences between intended access design and real usage. This assessment is grounded in what is actually happening, not in what policies assume should happen.

Third, the platform supports remediation by integrating with existing identity and security processes. Teams can rank risks by potential impact, assign ownership to the appropriate control teams, and monitor progress as issues are addressed. The goal is coordination across current controls rather than replacement.

Finally, because discovery and analysis operate continuously, evidence for governance and compliance is available at all times. Current application inventories, records of identity usage, and documentation of control gaps and corrective actions are maintained on an ongoing basis. This shifts audits from periodic, manual exercises to a continuous readiness model.

As identity increasingly moves into application layers, sustained visibility into how access actually functions becomes essential for reducing unmanaged exposure, improving audit preparedness, and enabling decisions based on verified operational data rather than assumptions.

Smart Homes Under Threat: How to Reduce the Risk of IoT Device Hacking

 

Most households today use some form of internet of things (IoT) technology, whether it’s a smartphone, tablet, smart plugs, or a network of cameras and sensors. Learning that nearly 120,000 home security cameras were compromised in South Korea and misused for sexploitation footage is enough to make anyone reconsider adding connected devices to their living space. After all, the home is meant to be a private and secure environment.

Although all smart homes carry some level of risk, widespread hacking incidents are still relatively uncommon. Cybercriminals targeting smart homes tend to be opportunistic rather than strategic. Instead of focusing on a particular household and attempting to break into a specific system, they scan broadly for devices with weak or misconfigured security settings that can be exploited easily.

The most effective way to safeguard smart home devices is to avoid being an easy target. Unfortunately, many of the hacking cases reported in the media stem from basic security oversights that could have been prevented with simple precautions.

How to Protect Your Smart Home From Hackers

Using weak passwords, neglecting firmware updates, or leaving Wi-Fi networks exposed can increase the risk of unauthorized access—even if the overall threat level remains low. Below are key steps homeowners can take to strengthen smart home security.

1. Use strong and unique passwords
Hackers gaining access to baby monitors and speaking through two-way audio is often the result of unchanged default passwords. Weak or reused passwords are easy to guess, especially if they have appeared in previous data breaches. Each smart device and account should have a strong, unique password to make attacks more difficult and less appealing.

2. Enable two-factor or multi-factor authentication
Multi-factor authentication adds an extra layer of protection by requiring a second form of verification beyond a password. Even if login credentials are compromised, attackers would still need additional approval. Many major smart home platforms, including Amazon, Google, and Philips Hue, support this feature. While it may add a small inconvenience during login, the added security is well worth the effort.

3. Secure your Wi-Fi network
Wi-Fi security is often overlooked but plays a critical role in smart home protection. Using WPA2 or WPA3 encryption and changing the router’s default password are essential steps. Limiting who has access to your Wi-Fi network also helps. Creating separate networks—one for personal devices and another exclusively for IoT devices—can further reduce risk by isolating smart home hardware from sensitive data.

4. Keep device firmware updated
Manufacturers regularly release firmware updates to patch newly discovered vulnerabilities. Enabling automatic updates ensures devices receive these fixes promptly. Keeping firmware current is one of the simplest and most effective ways to close security gaps.

5. Disable unnecessary features
Features that aren’t actively used can create additional entry points for attackers. If remote access isn’t needed, disabling it can significantly reduce exposure—particularly for devices with cameras. It’s also advisable to turn off Universal Plug and Play (UPnP) on routers and decline unnecessary integrations or permissions that don’t serve a clear purpose.

6. Research brands before buying
Brand recognition alone doesn’t guarantee strong security. Even well-known companies such as Wyze, Eufy, and Google have faced security issues in the past. Before purchasing a smart device, it’s important to research the brand’s security practices, data protection policies, and real-world user experiences. If features like local-only storage are important, they should be verified through reviews, forums, and independent evaluations.

Smart homes offer convenience and efficiency, but they also demand responsibility. By following basic cybersecurity practices and making informed purchasing decisions, homeowners can significantly reduce risks and enjoy the benefits of connected living with greater peace of mind.

China-Linked Hackers Step Up Quiet Spying Across South-East Asia

Threat actors linked to China have been blamed for a new wave of cyber-espionage campaigns targeting government and law-enforcement agencies across South-East Asia during 2025, according several media reports. Researchers at Check Point Research said they are tracking a previously undocumented cluster, which they have named Amaranth-Dragon, that has targeted Cambodia, Thailand, Laos, Indonesia, Singapore and the Philippines. 

The activity shows technical and operational links to APT41, a well-known Chinese hacking ecosystem.  
“Many of the campaigns were timed to coincide with sensitive local political developments, official government decisions, or regional security events,” Check Point said. “By anchoring malicious activity in familiar, timely contexts, the attackers significantly increased the likelihood that targets would engage with the content.” 

The firm described the operations as tightly scoped and deliberately restrained, suggesting an effort to establish long-term access rather than cause disruption. Infrastructure was configured to communicate only with victims in specific countries, reducing the risk of discovery. 

A key technique involved exploiting CVE-2025-8088, a now-patched flaw in WinRAR that allows arbitrary code execution when a malicious archive is opened. Check Point said the group began exploiting the vulnerability within days of its public disclosure in August. “The speed and confidence with which this vulnerability was operationalised underscores the group’s technical maturity and preparedness,” the researchers said. 

Although the initial infection vector remains unclear, analysts believe spear-phishing emails were used to distribute malicious RAR files hosted on cloud services such as Dropbox. Once opened, the archive launches a loader using DLL side-loading, a tactic frequently associated with Chinese groups. The loader then retrieves an encryption key from one server, decrypts a payload from another location and executes it directly in memory. 

The final stage deploys Havoc, an open-source command-and-control framework. Earlier versions of the campaign relied on ZIP files containing Windows shortcuts and batch files, while a separate operation in Indonesia delivered a custom remote-access trojan known as TGAmaranth RAT. That malware used a hard-coded Telegram bot for command and control and supported functions such as taking screenshots, running shell commands and transferring files. 

Check Point said the command infrastructure was shielded by Cloudflare and restricted by geography, accepting traffic only from targeted countries. Compilation times and working patterns pointed to operators based in China’s time zone. 

“In addition, the development style closely mirrors established APT41 practices,” the company said, adding that overlaps in tools and techniques suggest shared resources within the ecosystem. The findings come as another Chinese group, Mustang Panda, was linked to a separate espionage campaign uncovered by Dream Research Labs. The operation, dubbed PlugX Diplomacy, targeted officials involved in diplomacy, elections and international coordination between December 2025 and mid-January 2026.  

“Rather than exploiting software vulnerabilities, the operation relied on impersonation and trust,” Dream said. 

Victims were lured into opening files disguised as diplomatic or policy documents, which triggered infection automatically. The files installed a modified version of PlugX, a long-used Chinese espionage tool, through a multi-step process involving Windows shortcuts, PowerShell scripts and DLL search-order hijacking using a legitimate signed executable. A decoy document was shown to victims while the malware quietly embedded itself in the system. 

“The correlation between actual diplomatic events and the timing of detected lures suggests that analogous campaigns are likely to persist as geopolitical developments unfold,” Dream concluded.

Experts Find Malicious Browser Extensions, Chrome, Safari, and Edge Affected


Threat actors exploit extensions

Cybersecurity experts found 17 extensions for Chrome, Edge, and Firefox browsers which track user's internet activity and install backdoors for access. The extensions were downloaded over 840,000 times. 

The campaign is not new. LayerX claimed that the campaign is part of GhostPoster, another campaign first found by Koi Security last year in December. Last year, researchers discovered 17 different extensions that were downloaded over 50,000 times and showed the same monitoring behaviour and deploying backdoors. 

Few extensions from the new batch were uploaded in 2020, exposing users to malware for years. The extensions appeared in places like the Edge store and later expanded to Firefox and Chrome. 

Few extensions stored malicious JavaScript code in the PNG logo. The code is a kind of instruction on downloading the main payload from a remote server. 

The main payload does multiple things. It can hijack affiliate links on famous e-commerce websites to steal money from content creators and influencers. “The malware watches for visits to major e-commerce platforms. When you click an affiliate link on Taobao or JD.com, the extension intercepts it. The original affiliate, whoever was supposed to earn a commission from your purchase, gets nothing. The malware operators get paid instead,” said Koi researchers. 

After that, it deploys Google Analytics tracking into every page that people open, and removes security headers from HTTP responses. 

In the end, it escapes CAPTCHA via three different ways, and deploy invisible iframes that do ad frauds, click frauds, and tracking. These iframes disappear after 15 seconds.

Besides this, all extensions were deleted from the repositories, but users shoul also remove them personally. 

This staged execution flow demonstrates a clear evolution toward longer dormancy, modularity, and resilience against both static and behavioral detection mechanisms,” said LayerX. 

The PNG steganography technique is employed by some. Some people download JavaScript directly and include it into each page you visit. Others employ bespoke ciphers to encode the C&C domains and use concealed eval() calls. The same assailant. identical servers. many methods of delivery. This appears to be testing several strategies to see which one gets the most installs, avoids detection the longest, and makes the most money.

This campaign reflects a deliberate shift toward patience and precision. By embedding malicious code in images, delaying execution, and rotating delivery techniques across identical infrastructure, the attackers test which methods evade detection longest. The strategy favors longevity and profit over speed, exposing how browser ecosystems remain vulnerable to quietly persistent threats.

Makina Finance Loses $4M in ETH After Flash Loan Price Manipulation Exploit

 

One moment it was operating normally - then suddenly, price feeds went haywire. About 1,299 ETH vanished during what looked like routine activity. That sum now exceeds four million dollars in value. The trigger? A flash loan attack targeting Makina Finance, built on Ethereum. Not a hack of code - but an economic twist inside the system. Security teams such as PeckShield traced moves across the DUSD–DUSDC liquidity pool. Borrowed funds flooded in, shifting valuations without breaking access rules. Prices bent under pressure from artificial trades. Afterward, profits drained off-chain. What stayed behind were distorted reserves and puzzled users. No stolen keys. No failed signatures. Just manipulation riding allowed functions too far. 

The exploit started, researchers say, with a $280 million flash loan taken in USDC. Of that amount, roughly $170 million went toward distorting data from the MachineShareOracle, which sets values for the targeted liquidity pool. With prices artificially raised, trades worth around $110 million passed through the system - leaving over 1,000 ETH missing afterward. What happened fits a known pattern: manipulating value via temporary shifts in market depth. Since Makina's setup depended on immediate price points, sudden influxes of borrowed funds were enough to warp them. Inserting capital, pushing valuations up, then pulling assets out while gains lasted exposed a flaw built into how prices are calculated.  

Even though the exploit worked, the hacker did not receive most of the stolen money. A different actor, an MEV builder, stepped in ahead during the draining transaction and took nearly all the ETH pulled out. According to PeckShield, this twist could make getting back the assets more likely. Yet, there has been no public word from Makina on whether they have reached out to - or even found - the MEV searcher responsible. 

After reviewing what happened, Makina explained the vulnerability only touched its DUSD–DUSDC Curve pool, leaving everything else untouched. Security measures kicked in across all Machines - its smart vault network - as checks continue into how deep the effects go. To stay safe, users putting liquidity in that specific pool got a heads-up to pull out whatever they had left. More details will come once the team learns more through their ongoing review. 

Not long ago, flash loan attacks started showing up more often in DeFi. By October, the Bunni exchange closed for good following one such incident - $8.4 million vanished fast. Its team said restarting safely would mean spending too much on checks and oversight. Just weeks before, another hit struck Shibarium, a layer-two system. That breach pulled out $2.4 million in value almost instantly. 

Even so, wider trends hint at slow progress. Chainalysis notes that losses tied to DeFi stayed modest in 2025, though value held in decentralized systems climbed back near earlier peaks. Despite lingering dangers from flash loans, safeguards within the space seem to be growing more resilient over time.

Microsoft Unveils Backdoor Scanner for Open-Weight AI Models

 

Microsoft has introduced a new lightweight scanner designed to detect hidden backdoors in open‑weight large language models (LLMs), aiming to boost trust in artificial intelligence systems. The tool, built by the company’s AI Security team, focuses on subtle behavioral patterns inside models to reliably flag tampering without generating many false outcomes. By targeting how specific trigger inputs change a model’s internal operations, Microsoft hopes to offer security teams a practical way to vet AI models before deployment.

The scanner is meant to address a growing problem in AI security: model poisoning and backdoored models that act as “sleeper agents.” In such attacks, threat actors manipulate model weights or training data so the model behaves normally in most scenarios, but switches to malicious or unexpected behavior when it encounters a carefully crafted trigger phrase or pattern. Because these triggers are narrowly defined, the backdoor often evades normal testing and quality checks, making detection difficult. Microsoft notes that both the model’s parameters and its surrounding code can be tampered with, but this tool focuses primarily on backdoors embedded directly into the model’s weights.

To detect these covert modifications, Microsoft’s scanner looks for three practical signals that indicate a poisoned model. First, when given a trigger prompt, compromised models tend to show a distinctive “double triangle” attention pattern, focusing heavily on the trigger itself and sharply reducing the randomness of their output. Second, backdoored LLMs often leak fragments of their own poisoning data, including trigger phrases, through memorization rather than generalization. Third, a single hidden backdoor may respond not just to one exact phrase, but to multiple “fuzzy” variations of that trigger, which the scanner can surface during analysis.

The detection workflow starts by extracting memorized content from the model, then analyzing that content to isolate suspicious substrings that could represent hidden triggers. Microsoft formalizes the three identified signals as loss functions, scores each candidate substring, and returns a ranked list of likely trigger phrases that might activate a backdoor. A key advantage is that the scanner does not require retraining the model or prior knowledge of the specific backdoor behavior, and it can operate across common GPT‑style architectures at scale. This makes it suitable for organizations evaluating open‑weight models obtained from third parties or public repositories.

However, the company stresses that the scanner is not a complete solution to all backdoor risks. It requires direct access to model files, so it cannot be used on proprietary, fully hosted models. It is also optimized for trigger‑based backdoors that produce deterministic outputs, meaning more subtle or probabilistic attacks may still evade detection. Microsoft positions the tool as an important step toward deployable backdoor detection and calls for broader collaboration across the AI security community to refine defenses. In parallel, the firm is expanding its Secure Development Lifecycle to address AI‑specific threats like prompt injection and data poisoning, acknowledging that modern AI systems introduce many new entry points for malicious inputs.

Foxit Publishes Security Patches for PDF Editor Cloud XSS Bugs


 

In response to findings that exposed weaknesses in the way user-supplied data was processed within interactive components, Foxit Software has issued a set of security fixes intended to address newly identified cross-site scripting vulnerabilities. 

Due to the flaws in Foxit PDF Editor Cloud and Foxit eSign, maliciously crafted input could be rendered in an unsafe manner in the user's browser, potentially allowing arbitrary JavaScript execution during authenticated sessions. 

The fundamental problem was an inconsistency in input validation and output encoding in some UI elements (most notably file attachment metadata and layer naming logic), which enabled attacker-controlled payloads to persist and be triggered during routine user interactions. 

Among these issues, the most important one, CVE-2026-1591, affected the File Attachments list and Layers panel of Foxit PDF Editor Cloud, thus emphasizing the importance of rigorously enforcing client-side trust boundaries in order to prevent the use of seemingly low-risk document features as attack vectors. 

These findings were supported by Foxit's confirmation that the identified weaknesses were related to a specific way in which certain client-side components handled untrusted input within a cloud environment. Affected functionality allowed for the processing of user-controlled values — specifically file attachment names and PDF layer identifiers — without sufficient validation or encoding prior to rendering in the browser. 

By injecting carefully constructed payloads into the application's HTML context, carefully constructed payloads could be executed upon the interaction between an authenticated user and the affected interface components. In response to these security deficiencies, Foxit published its latest security updates, which it described as routine security and stability enhancements that require no remediation other than ensuring deployments are up to date. 

The advisory also identifies two vulnerabilities, tracked as CVE-2026-1591 and CVE-2026-1592, which are both classified under CWE-79 for cross-site scripting vulnerabilities. Each vulnerability has a CVSS v3.0 score of 6.3 and is rated Moderate in severity according to the advisory. 

Foxit PDF Editor Cloud is impacted by CVE-2026-1591, which has a significant impact on its File Attachments and Layers panels due to insufficient input validation and improper output encoding which can allow arbitrary JavaScript execution from the browser. 

The vulnerability CVE-2026-1592 poses a comparable risk through similar paths to data handling. Both vulnerabilities were identified and responsibly disclosed by Novee, a security researcher. However, the potential consequences of exploitation are not trivial, even if user interaction is required. In order to inject a script into a trusted browser context, an attacker would have to persuade a logged-in user to open or interact with a specially crafted attachment or altered layer configuration. 

By executing this script, an attacker can hijack a session, obtain unauthorized access to sensitive document data, or redirect the user to an attacker-controlled resource. As a result, the client-side trust assumptions made by document collaboration platforms pose a broader risk, particularly where dynamic document metadata is not rigorously sanitized. 

During the disclosure period, the source material did not enumerate specific CVE identifiers for each individual flaw, apart from those referenced in the advisory. The vulnerability involved in cross-site scripting has been extensively documented across a wide array of web-based applications and is routinely cataloged in public vulnerability databases such as MITRE's CVE repository.

XSS vulnerabilities in unrelated platforms, such as those described in CVE-2023-38545 and CVE-2023-38546, underscore the broader mechanics and effects of this attack category. This type of example is not directly related to Foxit products, but nevertheless is useful for gaining an understanding of how similar weaknesses may be exploited when web-rendered interfaces mishandle user-controlled data. 


Technically, Foxit PDF Editor Cloud is exploitable via the way it ingests, stores, and renders user-supplied metadata within interactive components like the File Attachments list and Layers dialog box. If input is not rigorously validated, an attacker may embed executable content (such as script tags or event handlers) into attachment filenames or layer names embedded within a PDF file without rigorous input validation. 

Upon presenting these values to the browser without appropriate output encoding, the application unintentionally enables the browser to interpret the injected content as active HTML or JavaScript as opposed to inert text. As soon as the malicious script has been rendered, it is executed within the security context of the authenticated user's session. 

The attacker can exploit the execution environment to gain access to session tokens and other sensitive browser information, manipulate the on-screen content, or redirect the user to unauthorized websites. Foxit cloud environments can be compromised with scripts that can perform unauthorized actions on behalf of users in more advanced scenarios. 

It is important to note that the risk is heightened by the low interaction threshold required to trigger exploitation, since simply opening or viewing a specially crafted document may trigger an injected payload, emphasizing the importance of robust client-side sanitization in cloud-based document platforms. 

These flaws are especially apparent in enterprise settings where Foxit PDF Editor Cloud is frequently integrated into day-to-day collaboration workflows. In such environments, employees exchange and modify documents sourced from customers, partners, and public repositories frequently, thereby increasing the risk that maliciously crafted PDFs could enter the ecosystem undetected. 

As part of its efforts to mitigate this broader risk, Foxit also publicly revealed and resolved a related cross-site scripting vulnerability in Foxit eSign, tracked as CVE-2025-66523, which was attributed to improper handling of URL parameters in specially constructed links. 

By enabling users to access these links with authenticated access, the untrusted input could be introduced into JavaScript code paths and HTML attributes without sufficient encoding, which could result in privilege escalation or cross-domain data exposure. A fix for this problem was released on January 15, 2026. 

Foxit confirmed that all identified vulnerabilities, including CVE-2026-1591, CVE-2026-1592, and CVE-2025-66523, have been fully addressed thanks to updates that strengthen both input validation and output encoding across all affected components. As a result of Foxit PDF Editor Cloud's automated updates or standard update mechanisms, customers are not required to perform any additional configuration changes. 

However, organizations are urged to verify that all instances are running the latest version of the application and remain alert for indicators such as unexpected JavaScript execution, anomalous editor behavior, or irregular entries in application logs which may indicate an attempt at exploitation.

Based on aggregate analysis, these issues are the result of a consistent breakdown in the platform's handling of user-controlled metadata during rendering of the File Attachments list and Layers panel. Insufficient validation controls allow attackers to introduce executable content through seemingly benign fields, such as attachment filenames or layer identifiers, through which malicious content may be introduced. This content, since it is not properly encoded, is interpreted by the browser as active code rather than plain text due to the lack of proper output encoding.

The injected JavaScript executes within the context of an authenticated session when triggered, resulting in a variety of outcomes, including data disclosure, interface manipulation, forced navigation, and unauthorised actions under the user's privilege. In addition to the low interaction threshold, the operational risks posed by these flaws are also highlighted by their limited access. 

While Foxit's remediation efforts address the immediate technical deficiencies, effective risk management extends beyond patch deployment alone. Organizations must ensure that all cloud-based instances are operating on current versions by applying updates promptly. 

In addition to these safeguards, other measures can be taken to minimize residual exposure, such as restricting document collaboration to trusted environments, enforcing browser content security policies, and monitoring application behavior for abnormal script execution.

Additional safeguards, such as web application firewalls and intrusion detection systems, are available at the perimeter of the network to prevent known injection patterns from reaching end users. Together with user education targeted at handling unsolicited documents and suspicious links, these measures can mitigate the broader threat posed by client-side injection vulnerabilities in collaborative documents.