Ivanti has disclosed two critical zero-day vulnerabilities, CVE-2026-1281 and CVE-2026-1340, in its Endpoint Manager Mobile (EPMM) product. Both flaws are being actively exploited and allow for unauthenticated remote code execution.
Vulnerability Details
Affected Product: Ivanti Endpoint Manager Mobile (EPMM) versions 12.5.0.0, 12.5.1.0, 12.6.0.0, 12.6.1.0, 12.7.0.0 and earlier.
Identifier: CVE-2026-1281, CVE-2026-1340
CVSS Score: 9.8 (Critical) for both vulnerabilities.
Exploitation Status: Actively Exploited. CVE-2026-1281 has been added to CISA's Known Exploited Vulnerabilities (KEV) catalog.
Risk & Impact
Triage: Immediate patching is required. Systems exposed to the internet should be considered compromised.
Attack Vector: The flaws are code injection vulnerabilities that allow a remote, unauthenticated attacker to execute arbitrary code. This can lead to complete system takeover, deployment of web shells, and lateral movement.
Ease of Exploit: Trivial; no authentication or user interaction is required for exploitation.
Action Plan
Immediate Action: Apply the version-specific RPM-based interim patches provided by Ivanti immediately. A permanent fix will be included in EPMM version 12.8.0.0, expected in Q1 2026.
Workaround: There is no workaround mentioned other than applying the patches. Note that the RPM patches do not persist after a version upgrade and must be reapplied.
Detection: Administrators should check Apache access logs for 404 HTTP response codes related to the URL path pattern containing "/mifs/c/" followed by either "aftstore" or "appstore". It is also advised to review for unauthorized configuration changes, new administrator accounts, and new push applications.
Relevant professional terms
Zero-Day Vulnerability
A security flaw in software that is unknown to the vendor. Attackers can exploit this vulnerability before a patch is available, giving the vendor "zero days" to fix it.
EPMM (Endpoint Manager Mobile)
A mobile management software solution, formerly known as MobileIron, that allows IT administrators to set policies and securely manage mobile devices, applications, and content within an organization.
Microsoft has released an emergency, out-of-band patch for a high-severity zero-day vulnerability, identified as CVE-2026-21509, that is being actively exploited. The flaw allows attackers to bypass security features in Microsoft Office by tricking a user into opening a specially crafted document, which can lead to malicious code execution.
Vulnerability Details
Affected Product: Microsoft Office 2016, 2019, LTSC 2021, LTSC 2024, and Microsoft 365 Apps for Enterprise.
Identifier: CVE-2026-21509
CVSS Score: 7.8 (High)
Exploitation Status: Actively Exploited
Risk & Impact
Triage: Immediate patching is strongly recommended due to active exploitation.
Attack Vector: The vulnerability is exploited via social engineering, where an attacker convinces a user to open a malicious Office document. This bypasses Object Linking and Embedding (OLE) mitigations designed to block unsafe controls, allowing malicious code to execute. The Preview Pane is not a known attack vector.
Ease of Exploit: The attack requires user interaction but is considered low complexity for the attacker once the malicious file is created.
Action Plan
Immediate Action: For Office 2021 and later, including Microsoft 365 Apps, close and restart all Office applications to apply the server-side fix. For Office 2016 and 2019, a manual update must be applied via Windows Update.
Workaround: No specific workarounds have been provided other than applying the patch. Organizations should ensure email gateway filters and endpoint protection are updated to help mitigate the threat.
Detection: Organizations should monitor for the execution of suspicious processes originating from Microsoft Office applications. Threat hunting should focus on identifying recently modified or created Office documents that trigger anomalous network activity or child processes.
Relevant professional terms
Zero-Day Vulnerability
A flaw in a computer system that is unknown to those who should be interested in mitigating it, including the vendor of the target software. Attackers can exploit it before a patch is available.
Malicious Code
Any computer programming script designed to create or exploit system vulnerabilities. It is used by threat actors to cause unwanted changes, damage systems, or gain unauthorized access.
The "Chat & Ask AI" application, which claims over 50 million users, exposed hundreds of millions of private user messages. The leak included highly sensitive conversations about topics such as suicide and illegal activities.
Attack Overview
Attack Path: The data exposure was reportedly due to a misconfigured Google Firebase database, discovered by an independent security researcher. This is indicative of insecure storage practices by the developer.
Impact Assessment
Data Stolen: Exposed data included entire chat histories, phone numbers, and email addresses from millions of users.
Strategic Takeaway
This incident highlights the critical failure to implement basic data security controls, emphasizing the need for robust security and privacy measures in the rapid development of consumer-facing AI applications.
Relevant professional terms
Data Leak
The unintentional exposure of confidential or sensitive information to an unauthorized party.
Misconfiguration
A security vulnerability that results from incorrectly setting up a system, such as a database, allowing for unauthorized public access.
Threat actors are exploiting the rising popularity of the AI assistant Moltbot, formerly Clawdbot, by distributing malicious VS Code extensions and skills. These attacks target developers and users of the open-source tool, aiming to steal credentials and establish backdoors by capitalizing on misconfigurations and brand recognition.
Key TTPs
Initial Access: Malicious VS Code extensions distributed via official marketplaces and poisoned "skills" in the public library.
Execution: Victims are tricked into running malicious commands, which can install Remote Access Trojans (RATs) like ScreenConnect.
Defense Evasion: Payloads are often hidden within legitimate-looking branding and Base64 encoded strings to avoid detection.
Campaign Analysis
The rebranding from Clawdbot to Moltbot created an opportunity for threat actors to hijack the brand's momentum. The campaign's impact is magnified by users granting the AI agent extensive system permissions, turning a misconfiguration into a critical vulnerability.
Targeting & Infrastructure
Target Profile: Developers and end-users installing the popular open-source AI assistant.
Infrastructure: Malicious code is hosted on third-party servers and distributed through the official VS Code Marketplace and Moltbot's skill repository.
Relevant Terms
Supply Chain Risk: The threat that a vulnerability in a third-party tool or library, like a poisoned AI "skill," can be used to attack downstream users.
Remote Access Trojan (RAT): Malware that provides an attacker with full remote control over an infected computer, enabling data theft and surveillance.
Unknown threat actors compromised the update infrastructure of eScan antivirus, using it to distribute multi-stage malware through legitimate update channels. The attack targeted both enterprise and consumer users, deploying a malicious payload designed to disable the antivirus software and establish persistence.
Key TTPs
Initial Access: Compromised a regional update server to push a trojanized update file (Reload.exe).
Execution: The malicious update drops and runs a downloader payload (CONSCTLX.exe).
Defense Evasion: Tampered with the system's HOSTS file and eScan's registry settings to block legitimate updates and communication.
Campaign Analysis
This supply chain attack is significant as it turns a security tool into a malware distribution vector, eroding trust in the software supply chain. The payload's design to disable the host antivirus prevents automatic remediation, requiring manual intervention from affected users.
Targeting & Infrastructure
Target Profile: Global enterprise and consumer users of the eScan antivirus product.
Infrastructure: Utilized at least one compromised regional update server belonging to MicroWorld Technologies.
Relevant Terms
Supply Chain Attack: A cyberattack that targets a trusted third-party vendor or software that has access to a victim's systems or data.
C2 (Command and Control): The server infrastructure that attackers use to send commands to and receive data from a compromised system.
A widespread malware campaign is abusing the Hugging Face AI platform to host thousands of malicious Android APKs. These variants are designed to steal credentials from popular financial and payment applications.
Key TTPs
Initial Access: Users are lured into installing a dropper app, disguised as a security tool, via scareware-style ads.
Execution: The dropper fakes a Google Play update to download the final payload from a Hugging Face repository.
Defense Evasion: Attackers use server-side polymorphism to generate new, unique payload variants approximately every 15 minutes to evade signature-based detection.
Campaign Analysis
This campaign highlights a trend of threat actors abusing trusted, legitimate cloud and AI platforms to bypass reputation-based security filters. The high rate of payload generation indicates a sophisticated and automated effort to maintain persistence and avoid detection.
Targeting & Infrastructure
Target Profile: Android users with popular financial and payment service apps, such as Alipay and WeChat.
Infrastructure: The campaign leverages Hugging Face's CDN for payload delivery and a command-and-control (C2) server for data exfiltration and redirection.
Actionable Intelligence
IPs: 154[.]198[.]48[.]57
Domains: trustbastion[.]com
Relevant Terms
Polymorphism: A technique used by malware to constantly change its own code or signature to avoid detection by antivirus software.
C2 (Command and Control): A server or infrastructure used by attackers to send commands to compromised devices and receive stolen data.
A study by Wiz Research and AI security lab Irregular reveals that AI agents can autonomously identify and exploit web vulnerabilities, demonstrating capabilities that can exceed human performance in specific, targeted scenarios.
Key Findi]ngs
In targeted challenges, AI agents successfully solved 9 out of 10 offensive security tasks.
AI performance declined in broader, more realistic scenarios, where costs increased by 2 to 2.5 times.
While AI agents excel at pattern matching and known attacks, they struggle with tasks requiring specialized tools or creative problem-solving compared to humans.
The Bottom Line
The era of AI-driven cyberattacks is here. While human expertise remains superior for complex, out-of-the-box thinking, AI provides attackers with unprecedented scale and speed for initial reconnaissance and exploiting known vulnerability patterns. Security leaders must assume adversaries will leverage AI to automate discovery, forcing a strategic shift towards AI-powered defensive mechanisms and continuous, automated security validation to keep pace with the evolving threat landscape.
Relevant Terms
AI Agent: A software program powered by artificial intelligence that can perceive its environment and take autonomous actions to achieve specific goals, such as identifying security flaws.
Attack Surface: The total number of all possible entry points for an unauthorized user to access or extract data from a system.
