The post Cloudflare Cuts 20% of Staff to Pivot Toward an “AI-First Agentic” Future appeared first on Daily CyberSecurity.

Dirty Frag: unpatched Linux kernel flaw grants root access on Ubuntu, RHEL and Fedora. A working exploit is already public.

Security researchers have disclosed a new unpatched vulnerability in the Linux kernel, code-named Dirty Frag, that allows an unprivileged local user to gain full root access on most major Linux distributions, including Ubuntu, RHEL, Fedora, AlmaLinux, and CentOS Stream.

Dirty Frag is related to the Dirty Pipe family of vulnerabilities but is independent of the Copy Fail mitigation, meaning systems that already applied the algif_aead blacklist remain fully exposed.

“[the flaw] can obtain root privileges on major Linux distributions by chaining the xfrm-ESP Page-Cache Write vulnerability and the RxRPC Page-Cache Write vulnerability.” reads the advisory. “Dirty Frag is a case that extends the bug class to which Dirty Pipe and Copy Fail belong. Because it is a deterministic logic bug that does not depend on a timing window, no race condition is required, the kernel does not panic when the exploit fails, and the success rate is very high.”

The researcher Hyunwoo Kim (@v4bel) first disclosed the vulnerability.

The vulnerability chains two separate flaws. The first is the xfrm-ESP Page-Cache Write bug, rooted in the Linux IPsec subsystem and introduced in a January 2017 source code commit, the same commit responsible for CVE-2022-27666, a buffer overflow affecting multiple Linux distributions. The second is the RxRPC Page-Cache Write bug, introduced in June 2023. Neither flaw alone is sufficient on all systems, but together they cover each other’s blind spots: where one path is blocked by the environment, such as Ubuntu’s AppArmor restrictions on namespace creation, the other opens. The chain is what makes Dirty Frag universally dangerous across distributions.

“What both vulnerabilities have in common is that, on a zero-copy send path where splice() plants a reference to a page cache page that the attacker only has read access to into the frag slot of the sender side skb as is, the receiver side kernel code performs in-place crypto on top of that frag.” reads the analysis. “As a result, the page cache of files that an unprivileged user only has read access to (such as /etc/passwd or /usr/bin/su) is modified in RAM, and every subsequent read sees the modified copy.”

What makes Dirty Frag particularly dangerous is its reliability. Unlike many kernel exploits that depend on precise timing windows or race conditions, this is a deterministic logic bug. It doesn’t panic the kernel on failure, and its success rate is described as very high. A working proof-of-concept is already public, reducing exploitation to a single command.

The disclosure itself was complicated: the embargo broke early after a third party published detailed technical information and the exploit code without coordination. No CVE identifier has been assigned yet.

“Chaining the two variants makes the blind spots cover each other. In an environment where user namespace creation is allowed, the ESP exploit runs first. Conversely, on Ubuntu where user namespace creation is blocked but rxrpc.ko is built, the RxRPC exploit works” concludes the report.

Until official patches are available, the recommended workaround is to blocklist the esp4, esp6, and rxrpc kernel modules to prevent them from loading.

Follow me on Twitter: @securityaffairs and Facebook and Mastodon

Pierluigi Paganini

(SecurityAffairs – hacking, Dirty Frag)

The Pentagon is integrating AI into military operations, transforming cybersecurity, targeting, and command systems into a unified warfare architecture.

May 2026 marks a turning point in the evolution of modern warfare: the convergence of artificial intelligence, cybersecurity, and conventional military power is no longer theoretical. It is becoming an operational reality.

The Pentagon has signed agreements with major technology companies, including OpenAI, Google, Microsoft, Amazon, and SpaceX to integrate advanced AI models into classified military networks. The stated goal is clear: transform the United States into an “AI-first” military force capable of maintaining decision superiority across every battlefield domain.

Under this strategy, AI is no longer treated as a laboratory tool or analytical assistant. It is moving directly into the military chain of command, intelligence analysis, logistics, targeting, and operational planning. More than 1.3 million Department of Defense employees are already using the GenAI.mil platform, dramatically reducing processes that once took months to just days.

The Pentagon’s doctrine reflects a major cultural shift: code and combat are no longer separate domains. Cybersecurity itself is now considered a combat capability. The ability to deploy, secure, update, and operate AI models inside classified environments has become part of national defense infrastructure.

The contracts signed with technology providers include “lawful operational use” clauses, requiring vendors to accept any use considered legitimate by the Pentagon, including autonomous weapons systems and intelligence operations. This raises profound ethical and geopolitical questions.

At the same time, the U.S. military is pushing for deep integration across defense systems. Through the Army’s new “Right to Integrate” initiative, manufacturers of missiles, drones, radars, and sensors are being asked to open their software interfaces so AI agents can connect systems in real time. The inspiration comes largely from Ukraine, where open APIs allowed rapid battlefield integration between drones, sensors, and fire-control systems.

However, this transformation creates a dangerous paradox: the same openness that enables speed and flexibility also expands the attack surface. Every API, cloud platform, and AI integration point can potentially become an entry point for sophisticated adversaries such as China, Russia, or state-sponsored APT groups.

A compromised AI-enabled military ecosystem could allow attackers to inject false sensor data, manipulate targeting systems, degrade drone communications, study operational decision patterns, or even hijack autonomous weapons platforms. In this context, software vulnerabilities and supply-chain weaknesses are no longer merely IT problems, they become military objectives.

Washington is also increasingly concerned about the cyber risks posed by advanced AI models themselves. According to reports, the White House is considering new oversight mechanisms for frontier AI systems capable of autonomously discovering software vulnerabilities or automating cyberattacks at scale. Officials fear that uncontrolled deployment of such models could lead to mass exploitation of critical infrastructure, financial systems, or global supply chains.

The strategic implications extend beyond military technology. Major cloud providers such as Amazon, Microsoft, and Google are gradually becoming part of the American defense architecture. Civilian digital infrastructure is evolving into a structural extension of military power.

This raises difficult questions for Europe and Italy. In a world where most cloud, AI, and cybersecurity infrastructures are controlled by American companies, what does technological sovereignty really mean? Sovereignty is no longer just about producing chips or funding startups. It is about controlling the digital infrastructure that supports national defense, determining who can update AI systems operating on classified networks, and deciding who sets the operational rules of software during crises.

The United States, Israel, and China are already integrating AI into military doctrine at high speed. Europe risks remaining trapped between regulation and technological dependence unless it develops its own industrial capabilities, operational autonomy, and independent evaluation frameworks.

The message coming from Washington is unmistakable: the future of strategic power will depend on who controls AI models, data, interfaces, and software-driven operational systems. In modern warfare, software has become a battlefield domain, and the speed of code deployment increasingly matters as much as firepower itself.

A more detailed analysis is available in Italian here.

Follow me on Twitter: @securityaffairs and Facebook and Mastodon

Pierluigi Paganini

(SecurityAffairs – hacking, AI)

Palo Alto says hackers exploited PAN-OS zero-day CVE-2026-0300 for weeks, gaining root access to exposed firewalls and hiding traces.

Palo Alto Networks warned that suspected state-sponsored hackers have been exploiting the critical PAN-OS zero-day CVE-2026-0300 for nearly a month. After exploiting the flaw, attackers deployed tunneling tools such as EarthWorm and ReverseSocks5, used stolen credentials to probe Active Directory, and deleted logs and other evidence to hide the intrusion.

“We are aware of only limited exploitation of CVE-2026-0300 at this time. Unit 42 is tracking CL-STA-1132, a cluster of likely state-sponsored threat activity exploiting CVE-2026-0300. The attacker behind this activity exploited CVE-2026-0300 to achieve unauthenticated remote code execution (RCE) in PAN-OS software. Upon successful exploitation, the attacker was able to inject shellcode into an nginx worker process.” reads the advisory by the cybersecurity vendor. “Post-exploitation activity includes deployment of publicly available tunneling tools (EarthWorm, ReverseSocks5), Active Directory enumeration using credentials likely obtained from the firewall, and the systematic destruction of logs and other evidence of compromise.”

EarthWorm has been used in past attacks associated with several China-linked threat actors, including , APT41, CL-STA-0046, and Volt Typhoon.

The flaw is a buffer overflow that allows unauthenticated remote code execution, especially when the User-ID portal is exposed to the internet.

“A buffer overflow vulnerability in the User-ID Authentication Portal (aka Captive Portal) service of Palo Alto Networks PAN-OS software allows an unauthenticated attacker to execute arbitrary code with root privileges on the PA-Series and VM-Series firewalls by sending specially crafted packets.” reads the advisory published by Palo Alto Networks. “The risk of this issue is greatly reduced if you secure access to the User-ID Authentication Portal per the best practice guidelines by restricting access to only trusted internal IP addresses.”

This week, Palo Alto Networks has warned that the critical PAN-OS vulnerability CVE-2026-0300 is actively exploited in the wild.

Below is the list of impacted products:

The cybersecurity vendor states that the issue doesn’t impact Prisma Access, Cloud NGFW and Panorama appliances.

Palo Alto Networks says the flaw is being exploited in a limited way, mainly against systems where the User-ID Authentication Portal is exposed to the public internet.

The flaw remains unpatched, with fixes expected from May 13, 2026. It affects PA-Series and VM-Series firewalls using the User-ID Authentication Portal. Palo Alto Networks notes risk is much lower for organizations that follow best practices, like limiting access to trusted internal networks only.

“Limited exploitation has been observed targeting Palo Alto Networks User-ID Authentication Portals that are exposed to untrusted IP addresses and/or the public internet.” concludes the advisory. “Customers following standard security best practices, such as restricting sensitive portals to trusted internal networks are at a greatly reduced risk.”

EarthWorm is an open-source tunneling tool written in C that works across Windows, Linux, macOS, and ARM/MIPS platforms. It acts as a SOCKS5 proxy and port-forwarding utility, enabling attackers to create covert communication channels, bypass network restrictions, and move laterally within compromised environments. Its features include forward and reverse SOCKS5 tunnels, port bridging, traffic forwarding, and multi-hop tunneling for protocols such as RDP and SSH. The tool has previously been linked to threat groups including Volt Typhoon and APT41.

ReverseSocks5 is another open-source networking tool designed to bypass firewalls and NAT protections by creating outbound connections from compromised systems to attacker-controlled servers. Once connected, it establishes a SOCKS5 proxy tunnel that allows remote access into the internal network. While commonly used by administrators for legitimate remote management, threat actors also abuse it for stealthy pivoting and post-compromise operations.

“The reliance of the attackers behind CL-STA-1132 on open-source tooling, rather than proprietary malware, minimized signature-based detection and facilitated seamless environment integration. This technical choice, combined with a disciplined operational cadence of intermittent interactive sessions over a multi-week period, intentionally remained below the behavioral thresholds of most automated alerting systems.” concludes Palo Alto Networks. “The lateral movement technique prioritized identity trust abuse over traditional network-layer pivoting, effectively reducing the attacker’s footprint. Consequently, this campaign demonstrates that operational restraint—specifically the use of non-persistent access windows—is a primary factor in maintaining long-term residency on edge infrastructure.”

Follow me on Twitter: @securityaffairs and Facebook and Mastodon

Pierluigi Paganini

(SecurityAffairs – hacking, PAN-OS)

The U.S. Cybersecurity and Infrastructure Security Agency (CISA) adds a flaw in Ivanti Endpoint Manager Mobile (EPMM) to its Known Exploited Vulnerabilities catalog

The U.S. Cybersecurity and Infrastructure Security Agency (CISA) added a flaw in the Ivanti Endpoint Manager Mobile (EPMM), tracked as CVE-2026-6973 (CVSS score of 7.1), to its Known Exploited Vulnerabilities (KEV) catalog.

Ivanti warns customers of a high‑severity zero‑day vulnerability, tracked as CVE‑2026‑6973, in Endpoint Manager Mobile that is already being exploited.

“At the time of disclosure, we are aware of very limited exploitation of CVE-2026-6973, which requires admin authentication for successful exploitation.” reads the advisory. “We are not aware of any customers being exploited by the other vulnerabilities disclosed today.”

The flaw, caused by improper input validation, allows attackers with admin privileges to execute arbitrary code on systems running EPMM 12.8.0.0 and earlier. Customers are urged to patch immediately to prevent compromise.

Ivanti EPMM 12.6.1.1, 12.7.0.1, and 12.8.0.1 address the vulnerability. The vulnerability doesn’t affect Ivanti Neurons for MDM, Ivanti’s cloud-based unified endpoint management solution, Ivanti EPM (a similarly named, but different product), Ivanti Sentry, or any other Ivanti products.

According to Binding Operational Directive (BOD) 22-01: Reducing the Significant Risk of Known Exploited Vulnerabilities, FCEB agencies have to address the identified vulnerabilities by the due date to protect their networks against attacks exploiting the flaws in the catalog.

Experts also recommend that private organizations review the Catalog and address the vulnerabilities in their infrastructure.

CISA orders federal agencies to fix the vulnerability by May 10, 2026.

Pierluigi Paganini

Follow me on Twitter: @securityaffairs and Facebook and Mastodon

(SecurityAffairs – hacking, US CISA Known Exploited Vulnerabilities catalog)

Cisco fixed several high‑severity flaws in its enterprise products, including SSRF bugs in Unity Connection that could enable code execution or service disruption.

Cisco released patches for multiple high‑severity vulnerabilities affecting its enterprise products. Successful exploitation could allow code execution, server‑side request forgery (SSRF), or denial‑of‑service attacks. Two notable flaws, CVE‑2026‑20034 and CVE‑2026‑20035, impact Cisco Unity Connection. Attackers can exploit them to trigger SSRF attacks.

“Multiple vulnerabilities in Cisco Unity Connection could allow a remote attacker to execute arbitrary code on or conduct server-side request forgery (SSRF) attacks through an affected device.” reads the advisory published by Cisco.

CVE‑2026‑20034 is a flaw in Cisco Unity Connection that allows an authenticated remote attacker to run arbitrary root‑level code on the device. The issue stems from improper validation of user input, letting an attacker send a crafted API request to fully compromise the system. Cisco has released fixes, and no workarounds exist.

“This vulnerability is due to insufficient validation of user-supplied input. An attacker could exploit this vulnerability by submitting a crafted API request.” reads the advisory. “A successful exploit could allow the attacker to execute arbitrary code as root, possibly resulting in the complete compromise of a targeted device. To exploit this vulnerability, the attacker must have valid user credentials on the affected device.”

CVE-2026-20035 flaw in Cisco Unity Connection Web Inbox UI allows an unauthenticated remote attacker to perform SSRF attacks. The issue comes from improper validation of certain HTTP requests. By sending a crafted request, an attacker could make the device send arbitrary network traffic on their behalf, potentially accessing internal services.

“A vulnerability in the web UI of Cisco Unity Connection Web Inbox could allow an unauthenticated, remote attacker to conduct SSRF attacks through an affected device.” reads the advisory.

“This vulnerability is due to improper input validation for specific HTTP requests. An attacker could exploit this vulnerability by sending a crafted HTTP request to an affected device. A successful exploit could allow the attacker to send arbitrary network requests that are sourced from the affected device.”

Below are the impacted releases:

Cisco PSIRT said it is not aware of any public reports or active malicious exploitation of these vulnerabilities.

Follow me on Twitter: @securityaffairs and Facebook and Mastodon

Pierluigi Paganini

(SecurityAffairs – hacking, Unity Connection)

A new banking trojan known as TCLBANKER has been quietly making rounds, and its delivery method is as clever as it is concerning. Attackers are using a trojanized version of a legitimate, digitally signed installer to slip malware onto victims’ machines without raising immediate suspicion.

The campaign, tracked as REF3076, bundles a malicious MSI installer inside a ZIP file and exploits the trust people place in recognizable software names.

The infection begins when a victim runs what appears to be a legitimate Logitech application installer. Inside the package, threat actors have weaponized the Logi AI Prompt Builder, abusing a technique called DLL sideloading to sneak a malicious file into the process. Once the application starts, it automatically loads the harmful DLL without the user ever knowing anything went wrong.

Analysts at Elastic Security Labs identified this new Brazilian banking trojan, assessing it to be a significant evolution of an older malware family known as MAVERICK and SORVEPOTEL. The campaign appears to be in its early stages, with developer artifacts and an incomplete phishing page suggesting the attackers are still actively building out their infrastructure.

TCLBANKER primarily targets users in Brazil, specifically those who visit banking, fintech, and cryptocurrency websites. The trojan monitors the victim’s browser in real time, watching for visits to any of 59 targeted financial domains.

Hackers Abuse Signed Logitech Installer

When a match is found, it opens a live connection to the attacker’s command server and puts the operator in full control.

The scope of potential damage goes well beyond simple credential theft. The malware can display fake full-screen overlays that look like real banking interfaces, freeze the apparent desktop to confuse victims, and kill the Task Manager to prevent users from ending the malicious process. It is a coordinated operation designed to make fraud feel seamless from the attacker’s side.

The attackers took care to make the infection chain look as normal as possible. The malicious ZIP file contains an MSI installer that mimics the legitimate Logi AI Prompt Builder, a real Flutter-based application.

When installed, the trojanized package drops a fake DLL called screen_retriever_plugin.dll, which masquerades as a genuine Flutter plugin and gets loaded automatically at startup.

The loader inside this DLL is packed with tricks to avoid detection. It checks whether the system is running inside a sandbox or virtual machine, verifies that the user’s default language is Brazilian Portuguese, and even measures timing to catch emulation frameworks that speed up sleep calls.

If anything seems off, the malware simply stops running without leaving obvious traces. This environment-gating approach means the payload only decrypts itself on real, qualifying machines.

Self-Spreading Worm Modules Amplify the Threat

What makes TCLBANKER particularly dangerous is not just what it does on a single machine, but how far it can spread from there. The malware comes with two worm modules designed to send itself to the victim’s contacts using channels those contacts already trust.

The first hijacks the victim’s active WhatsApp Web session in the browser, silently messaging Brazilian contacts with a link to download the malware. The second abuses Microsoft Outlook through automation, sending phishing emails directly from the victim’s own email account.

Because these messages come from real, known senders, they are far harder for security filters to catch. The Outlook bot first harvests the victim’s contact list, then sends targeted emails that look completely authentic.

Elastic researchers noted that all command and file-serving infrastructure runs on Cloudflare Workers under a single account, making it easy for operators to rotate infrastructure quickly when needed.

Organizations and individuals can take several steps to reduce exposure. Keeping security software updated ensures the latest detection signatures are in place.

Being cautious about ZIP files or MSI installers received through messaging apps or email, even from known contacts, is critical given this trojan’s self-spreading behavior. Monitoring for unusual scheduled tasks, unexpected DLL loads alongside legitimate software, and suspicious outbound connections can also help flag infections early.

Indicators of Compromise (IoCs):-

Note: IP addresses and domains are intentionally defanged (e.g., [.]) to prevent accidental resolution or hyperlinking. Re-fang only within controlled threat intelligence platforms such as MISP, VirusTotal, or your SIEM.

Follow us on Google News, LinkedIn, and X to Get More Instant Updates, Set CSN as a Preferred Source in Google.

The post Hackers Abuse Signed Logitech Installer to Deploy TCLBANKER Banking Trojan appeared first on Cyber Security News.

A new open-source cybersecurity platform called DarkMoon has emerged as a significant advancement in autonomous penetration testing.

It provides security teams and DevSecOps professionals with a fully AI-powered vulnerability assessment system. DarkMoon integrates over 50 specialized offensive security tools, all managed through a controlled execution interface.

DarkMoon is an automated penetration testing platform that uses artificial intelligence to orchestrate complete security assessments without manual intervention.

Unlike traditional vulnerability scanners, DarkMoon deploys a multi-agent AI architecture where specialized sub-agents reason, plan, and execute real offensive security operations through a controlled Model Context Protocol (MCP) interface, a gatekeeper layer that ensures the AI never directly touches the underlying system.

The platform aligns with recognized security frameworks, including ISO 27001, NIST SP 800-115, and the MITRE ATT&CK methodology, making it a standards-compliant option for organizations seeking repeatable, evidence-based assessments.

DarkMoon AI-Powered Platform

When a target is provided via the command line, DarkMoon automatically progresses through a multi-phase assessment: discovering open ports and services, fingerprinting the technology stack, modeling the attack surface, and then deploying specialized sub-agents based on what it detects.

The platform dynamically triggers agents tailored to discovered technologies:

• CMS Agent — activates for WordPress, Drupal, Joomla, Magento, and Moodle environments
• Stack-Specific Agent — targets PHP, Node.js, Flask, ASP.NET, Spring Boot, and Ruby on Rails
• Active Directory Agent — covers NetExec, BloodHound, and 30+ Impacket scripts
• Kubernetes Agent — uses kubectl, Kubescape, and Kubeletctl
• GraphQL Agent — handles GraphQL-specific attack surfaces
• Headless Browser Agent — deployed when browser rendering is required

Multiple agents can execute in parallel across a hybrid infrastructure, significantly accelerating assessment timelines compared to sequential manual testing.

DarkMoon ships with a purpose-built Docker image housing over 50 compiled security tools organized by category.

Port scanning is handled by Naabu and Masscan; web application testing leverages Nuclei, ffuf, sqlmap, Arjun, and wafw00f; reconnaissance uses Subfinder, Katana, Waybackurls, and httpx; CMS testing relies on WPScan and CMSeeK; and network enumeration employs Hydra, dig, and SNMP tooling.

All tools are accessible inside the Docker toolbox without path configuration — the AI reasons and plans, the MCP controls execution, and the Docker container runs the tools in isolation.

DarkMoon is designed for security teams running continuous automated testing, DevSecOps engineers integrating security into CI/CD pipelines, bug bounty hunters accelerating target analysis, and security researchers exploring adaptive attack surfaces in real time.

The platform supports bug bounty mode natively, with command-line flags such as FOCUS, EXCLUDE, SEVERITY, and FORMAT=h1 interpreted directly by the AI agent.

DarkMoon is available on GitHub at github.com/ASCIT31/Dark-Moon and requires only Docker, Docker Compose, and an LLM API key from providers such as Anthropic, OpenAI, or OpenRouter with local model support via Ollama and llama.cpp also available.

The platform represents a broader industry trend toward autonomous AI-driven penetration testing that scales beyond the limits of human-only security teams.

Cybercriminals now enter through your suppliers instead of your front door – Free Webinar

The post DarkMoon AI-Powered Autonomous Penetration Testing Platform With 50+ Tools appeared first on Cyber Security News.

Trellix, the global cybersecurity firm formed from the merger of McAfee Enterprise and FireEye, has confirmed unauthorized access to a portion of its source code repository, with the RansomHouse ransomware group formally claiming responsibility for the attack.

Trellix reported a data breach involving unauthorized access to a portion of its source code repository, which was disclosed publicly around May 2, 2026.

Upon discovering the intrusion, Trellix immediately engaged leading forensic experts to investigate and has notified law enforcement authorities.

In an official statement published on its website, the company said: “Based on our investigation to date, we have found no evidence that our source code release or distribution process was affected, or that our source code has been exploited”.

The RansomHouse ransomware group formally named Trellix on its dark web leak site, claiming the compromise occurred on April 17, 2026.

The group published multiple screenshots reportedly demonstrating access to Trellix’s internal services and management dashboards, though they have not specified the volume of data exfiltrated or its nature.

Notably, RansomHouse listed the breach status as “Evidence Depends on You,” a hallmark tactic used to pressure victims into negotiations before releasing stolen data publicly.

RansomHouse is a sophisticated ransomware-as-a-service (RaaS) group known for deploying a unique ransomware variant called Mario ESXi, whose code shares lineage with the leaked Babuk ransomware source code, alongside a tool called MrAgent to target both Windows and Linux-based virtualized environments.

The group typically targets VMware ESXi infrastructure and exploits weak domain credentials and monitoring systems to gain privileged access.

RansomHouse distinguishes itself by positioning itself as a “professional mediator community,” often seeking payment for data deletion rather than decryption.

The full extent of the data exposure remains unspecified, and Trellix has not confirmed whether corporate or customer data beyond source code was accessed.

Preliminary investigations indicate no evidence that the software distribution pipeline or customer-facing products were tampered with.

The incident highlights the growing trend of ransomware groups targeting cybersecurity vendors themselves, organizations whose proprietary source code, if weaponized, could have far-reaching consequences for enterprise defenses globally.

Cybercriminals now enter through your suppliers instead of your front door – Free Webinar

The post Trellix Breach – RansomHouse Claims Access to Parts of Source Code appeared first on Cyber Security News.

A sophisticated new malware framework called PCPJack has been found actively targeting cloud environments across the internet, hunting for exposed services and stripping away credentials at scale.

The worm zeroes in on Docker, Kubernetes, Redis, and MongoDB deployments, turning misconfigured or vulnerable systems into footholds for credential theft and financial fraud. What sets it apart from most cloud-targeting malware is its unusual decision to skip cryptocurrency mining entirely, suggesting the operators are focused on a different kind of profit.

PCPJack starts its infection chain with a shell script called bootstrap.sh, which runs quietly on Linux-based cloud systems. That script prepares the environment, installs Python, downloads six specialized modules, sets up persistence, and launches the main orchestrator.

One of its first actions is to scan for and actively remove all traces of a rival threat group called TeamPCP, essentially taking over compromised machines that someone else had already infected, making it unusually competitive among cloud threat actors.

Researchers at SentinelOne identified PCPJack as a credential theft framework with worm-like spreading capabilities. According to SentinelOne security researcher Alex Delamotte, the toolset “harvests credentials from cloud, container, developer, productivity, and financial services, then exfiltrates the data through attacker-controlled infrastructure while attempting to spread to additional hosts.”

The research team believes the actor behind PCPJack may be a former TeamPCP member who left the group and started their own separate operation, given the technical overlap found between both campaigns.

The malware collects an unusually wide range of secrets, including SSH keys, Slack tokens, WordPress database credentials, OpenAI and Anthropic API keys, cloud provider tokens, and cryptocurrency wallet files.

It then encrypts all stolen data using X25519 ECDH and ChaCha20-Poly1305 before sending it to a Telegram channel, broken into small chunks to comply with message size limits. The attacker even tracks whether their cleanup of TeamPCP infections was successful, signaling deliberate and targeted competitive intent rather than opportunistic attack behavior.

PCPJack’s Worm-Like Propagation and CVE Exploitation

PCPJack spreads by actively scanning external cloud infrastructure for exposed services including Docker, Kubernetes, Redis, MongoDB, and RayML. The worm downloads hostname data from Common Crawl parquet files and uses them as scanning targets, letting it discover new victims without hardcoding any addresses directly into the code.

This design allows the attacker to cover up to 104 million potential entries during each cycle without requiring centralised coordination.

The worm exploits five publicly known vulnerabilities to break into new systems. These include CVE-2025-29927, an authentication bypass in Next.js middleware; CVE-2025-55182, a server-side deserialization flaw in React and Next.js known as “React2Shell”; CVE-2026-1357, an unauthenticated file upload vulnerability in WPVivid Backup; CVE-2025-9501, a PHP injection flaw in W3 Total Cache; and CVE-2025-48703, a shell injection issue in CentOS Web Panel.

Once inside, the worm harvests SSH keys and moves laterally by enumerating Kubernetes clusters and Docker daemons, then replicating itself to every reachable host.

Sliver Backdoor and Enterprise-Wide Credential Targeting

SentinelOne’s analysis also uncovered a Sliver-based backdoor on the attacker’s staging server, compiled in three variants to support x86_64, x86, and ARM system architectures. This backdoor grants the operator persistent remote access even after initial exploitation ends.

The binaries are saved locally as update.bin, update-386.bin, and update-arm.bin, designed to blend in with legitimate system maintenance file names to avoid immediately raising suspicion.

Beyond cloud infrastructure, PCPJack also targets messaging platforms, financial services, and enterprise productivity tools. The malware scans for credentials tied to services like Discord, DigitalOcean, Grafana Cloud, Google API, HashiCorp Vault, and 1Password, expanding potential damage far beyond a single environment. This wide reach points toward extortion, spam campaigns, and credential resale as the most likely endgame.

To reduce exposure, security teams should enforce multi-factor authentication across all cloud accounts and services. Using IMDSv2 in AWS environments is recommended to prevent metadata theft, and proper authentication must be enforced for Docker and Kubernetes API endpoints.

Organisations should follow least-privilege principles, avoid storing secrets in plaintext, and regularly audit environment variables and configuration files for sensitive data.

Indicators of Compromise (IoCs):-

Note: IP addresses and domains are intentionally defanged (e.g., [.]) to prevent accidental resolution or hyperlinking. Re-fang only within controlled threat intelligence platforms such as MISP, VirusTotal, or your SIEM.

Follow us on Google News, LinkedIn, and X to Get More Instant Updates, Set CSN as a Preferred Source in Google.

The post New PCPJack Worm Targets Docker, Kubernetes, Redis, and MongoDB for Credential Theft appeared first on Cyber Security News.

A new and evolving threat has caught the attention of cybersecurity researchers worldwide. A Windows-based information stealer known as NWHStealer has resurfaced with a more sophisticated delivery chain, now using the Bun JavaScript runtime as part of its infection process.

This shift makes it clear that the attackers behind this campaign are actively experimenting with lesser-known tools to stay ahead of security defenses.

NWHStealer is a Rust-based malware capable of stealing sensitive data from infected Windows systems. It spreads through Node.js scripts, MSI installers, and fake software downloads hosted on trusted platforms such as GitHub, GitLab, SourceForge, and Itch.io. Since it blends into legitimate-looking software packages, many users unknowingly download and run it without any suspicion.

Analysts at Malwarebytes identified the new delivery method during routine threat hunting activities.

Researcher Gabriele Orini noted that attackers have now incorporated Bun, a modern JavaScript toolkit built as a high-performance alternative to Node.js, into the malware’s delivery chain. Its relative newness in security circles makes it particularly appealing to attackers trying to slip past detection.

Once inside a system, NWHStealer is highly capable. It collects system information, steals saved browser data and passwords, drains cryptocurrency wallets, and targets applications like Discord, Steam, and FTP clients such as FileZilla.

It can also inject malicious code into browser processes, bypass Windows User Account Control, persist through scheduled tasks, and pull new command-and-control addresses from Telegram to keep the operation alive after partial takedowns.

The scale of this campaign is notable. Attackers continue to create fresh profiles on legitimate platforms to push new lures, making it difficult for moderators to respond quickly. The combination of data theft, persistence, and self-updating infrastructure makes NWHStealer a serious threat to both everyday users and organizations.

Bun Loader, Anti-VM Checks, and Encrypted C2

The infection begins with a ZIP archive disguised as a game trainer, software crack, or utility tool. Detected archive names include MOUSE_PI_Trainer_v1.0.zip, FiveM Mod.zip, TradingView-Activation-Script-0.9.zip, and AutoTune 2026.zip.

Inside sits Installer.exe, which carries JavaScript code bundled with the Bun runtime hidden within its .bun section.

The malicious JavaScript is divided into two key files. The first, sysreq.js, runs PowerShell and WMI commands to check whether the system is a real machine or a virtual one. It inspects CPU count, disk space, screen resolution, hardware manufacturers, and even the username, using a scoring system to decide whether to proceed with infection or stop entirely. This anti-VM layer is designed to avoid detection in automated security analysis environments.

The second file, memload.js, handles communication with the attacker’s command-and-control server. Strings and configurations are encrypted using XOR combined with base64 encoding, making static analysis much harder. The loader sends a report containing the victim’s public IP, system details, and a screenshot to the C2, then fetches an AES-encrypted payload and deploys NWHStealer directly into memory with minimal traces on disk.

Some analyzed ZIP files also include a secondary loader called dw.exe inside a folder labeled “DW.” A Readme.txt inside the archive tells users to run dw.exe manually if the main installer fails, giving attackers a fallback option if the primary C2 server goes offline. This dual-loader setup reflects a deliberate backup plan to ensure delivery regardless of temporary disruptions.

Staying Safe From NWHStealer

Given how widely this stealer is distributed, users should take practical steps to protect themselves. Only download software from official, verified sources and avoid file-sharing platforms unless the publisher’s identity and reputation are clearly established.

Always check a file’s digital signature before running it, as legitimate software will carry consistent, verifiable signing details.

It is also worth inspecting any downloaded archive before opening it. Malicious archives often have unusual file structures, mismatched content, or naming patterns that do not match what was advertised.

Staying cautious with downloads that seem too good to be true, whether a game cheat, a software activator, or a free tool, remains one of the most effective defenses against threats like NWHStealer.

Indicators of Compromise (IoCs):-

Note: IP addresses and domains are intentionally defanged (e.g., [.]) to prevent accidental resolution or hyperlinking. Re-fang only within controlled threat intelligence platforms such as MISP, VirusTotal, or your SIEM.

Follow us on Google News, LinkedIn, and X to Get More Instant Updates, Set CSN as a Preferred Source in Google.

The post New NWHStealer Delivery Chain Uses Bun Loader, Anti-VM Checks, and Encrypted C2 appeared first on Cyber Security News.

Mozilla has fixed a total of 423 Firefox security bugs in April 2026 alone, a figure nearly 20 times higher than its monthly average of about 21 bugs throughout 2025, driven by a groundbreaking agentic AI pipeline built around Anthropic’s Claude Mythos Preview and other large language models.

The surge was triggered by Mozilla’s early access to Claude Mythos Preview, which identified 271 of the 423 vulnerabilities fixed in April.

These were primarily shipped as part of Firefox 150, released on April 21, 2026, with additional fixes flowing into Firefox 149.0.2, 150.0.1, and 150.0.2. Of the 271 bugs attributed to Claude Mythos Preview in Firefox 150, 180 were rated sec-high, 80 were sec-moderate, and 11 were sec-low, meaning most were vulnerabilities exploitable via normal user behavior, such as simply visiting a malicious webpage.

Mozilla Patches 423 Firefox 0-Day

Beyond the 271 AI-identified bugs, the remaining 152 fixes included 41 externally reported bugs and 111 discovered through internal techniques, split roughly equally between Claude Mythos fixes shipped in other releases, bugs found with other AI models, and conventional fuzzing.

Anthropic’s own Frontier Red Team was separately credited with three standalone CVEs: CVE-2026-6746, CVE-2026-6757, and CVE-2026-6758.

Mozilla publicly disclosed 12 representative bug reports to demonstrate the depth of AI analysis.

These include a 15-year-old flaw in the <legend> HTML element (Bug 2024437), triggered by meticulous orchestration of recursion stack depths and cycle collection edge cases, and a 20-year-old use-after-free (UAF) in Firefox’s XSLT engine (Bug 2025977) where reentrant key() calls caused a hash table to free its backing store while a raw pointer remained in use.

Several bugs represent critical sandbox escape primitives, including a race condition over IPC allowing a compromised content process to manipulate IndexedDB refcounts to trigger a UAF (Bug 2021894), and a raw NaN crossing an IPC boundary masquerading as a tagged JavaScript object pointer to achieve a parent-process fake-object primitive (Bug 2022034).

One exploit even simulates a malicious DNS server by intercepting glibc function calls to trigger a buffer over-read during HTTPS Record and ECH parsing (Bug 2023958).

These sandbox escape bugs are notoriously difficult to surface via traditional fuzzing methods, making AI coverage particularly valuable for this attack surface.

Mozilla’s approach evolved from early static-analysis experiments using GPT-4 and Claude Sonnet 3.5, which produced too many false positives to be practical.

The breakthrough came with agentic harness systems that not only generate bug hypotheses but also create reproducible proof-of-concept test cases to dynamically validate them. This eliminated speculative false positives and made large-scale deployment feasible.

The pipeline was built atop Mozilla’s existing fuzzing infrastructure and parallelized across multiple ephemeral virtual machines, each assigned to hunt for vulnerabilities within a specific target file.

Mozilla integrated the full security bug lifecycle into the system: deduplication against known issues, triage, patch tracking, and release management.

Over 100 contributors worked to review, test, and ship the resulting patches, a testament to the sustained operational scale required.

Key Vulnerability Breakdown

Equally notable is what the AI pipeline failed to exploit, not due to limitation, but because of effective prior hardening.

Audit logs revealed numerous AI-driven attempts to exploit prototype pollution for sandbox escapes, all blocked by Mozilla’s earlier architectural decision to freeze JavaScript prototypes by default. This provided direct, measurable validation of previously shipped defense-in-depth mitigations.

Mozilla’s guidance is direct: any software project can begin using an agentic harness with a modern model today.

The initial prompts can be simple, essentially directing the model to find a bug in a specific code region and build a test case, with iteration improving effectiveness over time.

Mozilla plans to integrate this pipeline into its continuous integration (CI) system to scan incoming patches as they land, extending coverage from file-based to patch-based scanning.

Cybercriminals now enter through your suppliers instead of your front door – Free Webinar

The post Mozilla Patches 423 Firefox Vulnerabilities with Claude Mythos and Other AI Models appeared first on Cyber Security News.

Spring Cloud Config provides crucial server-side and client-side support for externalized configuration in distributed systems.

Recently, the Spring development team disclosed four security vulnerabilities impacting the Spring Cloud Config Server.

These flaws range from medium to critical severity, exposing environments to unauthorized arbitrary file access, cloud secrets leakage, and logging misconfigurations.

Because centralized configuration servers often hold sensitive keys for an entire microservice architecture, system administrators must immediately review and patch their infrastructure.

Spring Cloud Vulnerabilities

Directory Traversal Vulnerabilities

The most severe issue is CVE-2026-40982, a critical directory traversal vulnerability affecting the platform.

The Spring Cloud Config module allows applications to serve both text and binary files over the network.

An attacker can exploit this module by sending a specially crafted URL to the server, thereby bypassing restricted directories and accessing arbitrary files on the host system.

Security researchers Swapnil Paliwal, the AxiomCode security team, August 829, and rash18mi responsibly identified and reported this critical flaw.

Target GCP Secrets and Git Directories

Two additional high-severity vulnerabilities threaten Spring Cloud Config deployments.

CVE-2026-40981 affects organizations that use Google Secrets Manager as the backend for their configuration server.

Malicious actors can craft specific requests to the config server, exposing sensitive secrets from unintended Google Cloud Platform projects.

Meanwhile, CVE-2026-41002 introduces a time-of-check-time-of-use attack surface.

This vulnerability specifically targets the server’s base directory used to clone Git repositories.

Threat actors can manipulate files during the cloning process due to this race condition.

Security researcher Yu Bao from PayPal received credit for discovering and reporting this Git-related vulnerability.

Trace Logging Exposes Sensitive Information

A medium-severity vulnerability (CVE-2026-41004) affects the server’s internal logging mechanisms.

When administrators enable trace logging, the system inadvertently writes sensitive information in plain text directly to the log files.

This misconfiguration could expose credentials or configuration secrets to unauthorized internal users who possess read access to the system logs.

All four vulnerabilities impact the same branches of the Spring Cloud Config ecosystem.

The affected release lines include 3.1.x, 4.1.x, 4.2.x, 4.3.x, and 5.0.x. Older, unsupported versions of the software also remain highly vulnerable to these exploits.

Users must upgrade immediately to secure their environments against potential compromise.

The Spring team has released patched versions across their different support tiers.

Open-source software users must upgrade to 4.3. x environments to version 4.3.3 and their 5.0. x environments to version 5.0.3.

Enterprise support customers have access to dedicated fixes in versions 3.1.14, 4.1.10, and 4.2.7.

If immediate patching is impossible for the GCP secrets vulnerability, administrators can implement a temporary configuration workaround.

By setting the spring.cloud.config.server.gcp-secret-manager.token-mandatory=true property, the server forces clients to send a valid token.

The system then verifies this token to ensure the client actually has legitimate access to the requested project secrets.

Cybercriminals now enter through your suppliers instead of your front door – Free Webinar

The post Critical Spring Vulnerabilities Expose Arbitrary Files and GCP Secrets appeared first on Cyber Security News.

Dirty Frag is a newly disclosed, CVE-pending Linux kernel local privilege escalation (LPE) vulnerability that chains two separate page-cache write flaws, the xfrm-ESP Page-Cache Write and the RxRPC Page-Cache Write, to achieve root access on virtually all major Linux distributions, with a public exploit already in the wild following an embargo break on May 7, 2026.

Dirty Frag belongs to the same vulnerability class as Dirty Pipe and Copy Fail (CVE-2026-31431), but targets the frag member of the kernel’s struct sk_buff rather than struct pipe_buffer.

Discovered and reported by security researcher Hyunwoo Kim (@v4bel), the vulnerability exploits the zero-copy send path where splice() plants a reference to a read-only page cache page, such as /etc/passwd or /usr/bin/su — into the frag slot of a sender-side skb.

Dirty Frag Linux Vulnerability

The receiver-side kernel code then performs in-place cryptographic operations directly on top of that frag, permanently modifying the page cache in RAM.

Every subsequent read to that file sees the corrupted version, even though the unprivileged attacker was granted only read access.

Unlike race-condition exploits, Dirty Frag is a deterministic logic bug that requires no timing window, does not panic the kernel on failure, and carries an extremely high success rate.

xfrm-ESP Page-Cache Write resides in esp_input(), the IPsec ESP receive path. When an skb is non-linear but lacks a frag list, the code skips the mandatory skb_cow_data() buffer allocation step and jumps directly to in-place AEAD decryption on the attacker-planted frag.

Using the XFRMA_REPLAY_ESN_VAL netlink attribute, the attacker can control both the location (file offset) and the value (4 bytes) of each store operation, enabling them to overwrite arbitrary bytes of /usr/bin/su‘s page cache with a static root-shell ELF 192 bytes written across 48 chunks of 4 bytes each.

Authentication failure (-EBADMSG) is returned afterward, but the page cache write has already persisted. This variant requires the ability to create a user namespace (unshare(CLONE_NEWUSER)).

RxRPC Page-Cache Write resides in rxkad_verify_packet_1(), which performs an in-place single-block pcbc(fcrypt) decryption on the first 8 bytes of the RxRPC payload.

Because skb_to_sgvec() converts the splice-pinned page cache page directly into the SGL, the attacker-controlled page becomes both src and dst.

The 8-byte store value is fcrypt_decrypt(C, K), where K is a freely specifiable session key registered via add_key("rxrpc", ...) — an operation requiring no privileges at all.

The attacker brute-forces K in user space until the desired plaintext (e.g., turning /etc/passwd line 1’s password field into an empty string) is produced, enabling PAM nullok authentication bypass.

Neither vulnerability alone covers all Linux environments:

• ESP variant: Available on most distros but requires user namespace creation — blocked on some Ubuntu configurations via AppArmor policy.
• RxRPC variant: No namespace privilege required, but rxrpc.ko is absent on most distros like RHEL 10.1 by default — yet ships and auto-loads on Ubuntu.

Chaining the two exploits closes both blind spots, achieving root on essentially every major distribution. The exploit first attempts the ESP path; if unshare(CLONE_NEWUSER) fails, it automatically falls back to the RxRPC path targeting /etc/passwd.

Affected Distributions and Kernel Versions

The ESP vulnerability has been present since commit cac2661c53f3 (January 2017), and the RxRPC flaw since 2dc334f1a63a (June 2023), giving the chain an effective window of approximately 9 years. Confirmed affected distributions include:

• Ubuntu 24.04.4 (kernel 6.17.0-23-generic)
• RHEL 10.1 (kernel 6.12.0-124.49.1.el10_1.x86_64)
• openSUSE Tumbleweed (kernel 7.0.2-1-default)
• CentOS Stream 10 (kernel 6.12.0-224.el10.x86_64)
• AlmaLinux 10 (kernel 6.12.0-124.52.3.el10_1.x86_64)
• Fedora 44 (kernel 6.19.14-300.fc44.x86_64)

The ESP variant patch using the SKBFL_SHARED_FRAG flag to ensure splice-pinned pages always route through skb_cow_data() — was merged into the netdev tree on May 7, 2026.

The final merged patch was based on a shared-frag approach submitted by Kuan-Ting Chen. The RxRPC patch, which adds || skb->data_len to the existing skb_cloned() gate to force isolation of non-linear skbs, remains unmerged upstream.

No CVE identifiers have been assigned for either flaw as of publication, due to the premature embargo break by an unrelated third party on May 7, 2026 .

Immediate Mitigation

Since distribution-level patches are not yet available, administrators should immediately disable the affected kernel modules using the following command:

bashsh -c "printf 'install esp4 /bin/false\ninstall esp6 /bin/false\ninstall rxrpc /bin/false\n' > /etc/modprobe.d/dirtyfrag.conf; rmmod esp4 esp6 rxrpc 2>/dev/null; true"

This blacklists and unloads the esp4, esp6, and rxrpc modules, disrupting IPsec and RxRPC functionality as a trade-off.

Systems that rely on IPsec VPN tunnels should weigh operational impact carefully before applying the workaround and prioritize applying distribution-backported kernel patches once available.

The complete technical write-up and PoC exploit code are available at the researcher’s GitHub repository.

Cybercriminals now enter through your suppliers instead of your front door – Free Webinar

The post Dirty Frag Linux Vulnerability Let Attackers Gain Root Privileges – PoC Released appeared first on Cyber Security News.

Vercel has released an extensive set of security advisories for Next.js, addressing more than a dozen vulnerabilities, including denial-of-service, middleware bypass, server-side request forgery, and cross-site scripting.

The flaws affect Next.js versions 13.x through 16.x using the App Router, as well as React Server Components packages for versions 19.x.

CVE-2026-23870: Denial of Service via React Server Components

A high-severity denial-of-service vulnerability tracked as CVE-2026-23870 affects React Server Components packages for versions 19.x and all Next.js App Router deployments on versions 13.x, 14.x, 15.x, and 16.x.

A specially crafted HTTP request sent to any App Router Server Function endpoint, when deserialized, can trigger excessive CPU usage, resulting in denial-of-service attacks in unpatched environments.

The issue is rooted in the React “Flight” protocol’s deserialization logic, which fails to adequately enforce structural or type constraints on inbound payloads.

Middleware and Proxy Authorization Bypass

Three separate advisories GHSA-267c-6grr-h53f, GHSA-26hh-7cqf-hhc6, and GHSA-492v-c6pp-mqqv address middleware bypass vulnerabilities in App Router applications.

Specially crafted .rsc and segment-prefetch URLs can resolve to the same page without being matched by intended middleware rules, allowing protected content to be accessed without proper authorization checks.

The fix now includes App Router transport variants when generating middleware matchers, ensuring middleware protections apply consistently to all request types, including prefetch variants.

Until an upgrade is possible, developers should enforce authorization directly in the underlying route or page logic rather than relying solely on middleware.

CVE-2026-44578: SSRF via WebSocket Upgrade Requests

Tracked as CVE-2026-44578 and covered under GHSA-c4j6-fc7j-m34r, this high-severity flaw enables server-side request forgery through crafted WebSocket upgrade requests on self-hosted Node.js deployments.

An attacker can manipulate the server into proxying requests to arbitrary internal or external destinations, potentially exposing internal services or cloud metadata endpoints, a particularly dangerous scenario in cloud-native environments.

Vercel-hosted deployments are explicitly noted as unaffected. The fix applies the same safety checks to WebSocket upgrade handling that already existed for standard HTTP requests.

CVE-2026-44573: Pages Router i18n Middleware Bypass

CVE-2026-44573 (GHSA-36qx-fr4f-26g5) affects applications using the Pages Router with i18n configured alongside middleware-based authorization.

Locale-less /_next/data/<buildId>/<page>.json requests bypass middleware entirely, enabling attackers to retrieve server-side rendered JSON for protected pages without passing authorization checks.

The matcher logic has been updated to apply consistent matching across both prefixed and unprefixed data routes.

Beyond the high-severity flaws, Vercel also patched several moderate and low-severity issues.

These include cross-site scripting vulnerabilities in App Router applications using CSP nonces (GHSA-ffhc-5mcf-pf4q) and in beforeInteractive scripts with untrusted input (GHSA-gx5p-jg67-6x7h), a denial-of-service bug in the Image Optimization API (GHSA-h64f-5h5j-jqjh), and cache poisoning issues in React Server Component responses (GHSA-wfc6-r584-vfw7, GHSA-vfv6-92ff-j949).

A connection exhaustion DoS in Cache Components (GHSA-mg66-mrh9-m8jx) and cache poisoning of middleware redirects (GHSA-3g8h-86w9-wvmq) round out the advisory list.

Organizations running affected Next.js versions should prioritize upgrading immediately.

For teams unable to upgrade right away, the recommended interim mitigations include enforcing authorization within individual route or page logic rather than relying on middleware alone, blocking WebSocket upgrades at the reverse proxy or load balancer level, and restricting server egress to known internal networks.

Cybercriminals now enter through your suppliers instead of your front door – Free Webinar

The post Multiple Critical Vulnerabilities Patched in Next.js and React Server Components appeared first on Cyber Security News.

Ivanti has issued a critical security advisory for its Endpoint Manager Mobile (EPMM) product, disclosing multiple actively exploited vulnerabilities, including CVE-2026-6973, and urging all on-premises EPMM customers to apply patches immediately.

At the time of disclosure, Ivanti confirmed active exploitation of CVE-2026-6973, a vulnerability that requires admin authentication to succeed.

The flaws exclusively affect the on-premises EPMM product and are not present in Ivanti Neurons for MDM, Ivanti’s cloud-based unified endpoint management solution, Ivanti EPM, Ivanti Sentry, or any other Ivanti products.

Exploitation activity has been described as “very limited” at the time of public disclosure, though the company strongly warned that advanced AI models have dramatically collapsed the time-to-exploit window from days to mere hours after a vulnerability becomes public.

In a notable shift in vulnerability management strategy, Ivanti disclosed that it has integrated multiple advanced large language model (LLM) AI systems into its product security and engineering red team processes.

This integration has enhanced the capabilities of its internal security teams to identify and remediate vulnerabilities that traditional static analysis (SAST) and dynamic analysis (DAST) tools typically miss.

Ivanti acknowledged that some of the vulnerabilities being disclosed today were discovered directly through this AI-assisted process. The company maintains a “human in the loop” policy to verify all automated or agentic findings, ensuring responsible use of AI in its security program.

Ivanti’s EPMM has been a recurring target for sophisticated threat actors. CISA has flagged at least 31 Ivanti defects on its Known Exploited Vulnerabilities (KEV) catalog since late 2021, and at least 19 defects across Ivanti products have been exploited in the past two years alone.

Previous zero-day campaigns against EPMM include CVE-2025-4427 and CVE-2025-4428 in May 2025, and CVE-2023-35078 and CVE-2023-35082 in 2023, with some attacks attributed to Chinese state-sponsored threat groups.

The consistent targeting of EPMM underscores the product’s high-value position in enterprise mobile device management infrastructure.

The vulnerabilities disclosed in Ivanti’s May 2026 security advisory affect only on-premises EPMM deployments. Organizations running cloud-based Ivanti Neurons for MDM are not impacted.

Ivanti has published detailed remediation instructions through its official Security Advisory, with patch packages that the company says take only seconds to apply and cause no downtime.

Mitigations

Ivanti strongly urges all on-premises EPMM administrators to take immediate action:

• Apply the available security patch to all EPMM on-premises instances without delay
• Monitor Apache access logs at /var/log/httpd/https-access_log for signs of attempted or successful exploitation.
• Implement network segmentation to restrict EPMM administrative interfaces to trusted networks only.
• Review and harden mobile device management policies to reduce the overall attack surface
• Subscribe to Ivanti’s Security Blog and the Ivanti Innovators Hub for real-time vulnerability alerts

Ivanti cautioned that as AI-driven tooling becomes further embedded in its security processes, customers should expect an increase in vulnerability disclosures, a transparency initiative the company frames as a proactive step toward more resilient products rather than a sign of weakening security posture.

Cybercriminals now enter through your suppliers instead of your front door – Free Webinar

The post New Ivanti EPMM 0-Day Vulnerability Actively Exploited in Attacks appeared first on Cyber Security News.

A sophisticated spear-phishing campaign, dubbed Operation GriefLure, targeting senior executives in Vietnam and the Philippines with a stealthy modular remote access trojan (RAT). The campaign focuses on high-value organizations, including Viettel Group Vietnam’s largest military-backed telecom provider and St. Luke’s Medical Center (SLMC) in the Philippines, demonstrating a calculated approach to regional cyber-espionage. What sets Operation […]

The post Modular RAT Campaign Steals Credentials and Captures Screenshots appeared first on GBHackers Security | #1 Globally Trusted Cyber Security News Platform.