Defender CVE-2026-41091 45498 Exploit Zero-Days Patched

Microsoft Defender Zero-Days CVE-2026-41091, 45498 Exploited

Microsoft disclosed two actively exploited zero-day vulnerabilities, CVE-2026-41091 and CVE-2026-45498, affecting the Microsoft Malware Protection Engine and Microsoft Defender Antimalware Platform. These flaws represent a significant threat to Windows systems globally, prompting immediate action from security professionals. The U.S. Cybersecurity and Infrastructure Security Agency (CISA) emphasized the urgency, directing federal agencies to patch these vulnerabilities by June 3, 2026.

The vulnerabilities allow for privilege escalation and denial-of-service, respectively, in widely deployed Microsoft antivirus and antispyware software. Their active exploitation necessitates prompt updates to mitigate potential system compromise across diverse environments. This incident shows ongoing challenges in securing foundational software components against sophisticated attack vectors.

Beyond the critical Defender updates, the past 24 hours in cybersecurity also included a significant supply chain attack against GitHub's internal repositories by TeamPCP, Microsoft's large-scale disruption of a malware-signing service, a data breach at 7-Eleven attributed to ShinyHunters, and a precautionary pause in U.S. bank cybersecurity examinations due to potential risks from Anthropic PBC's new Mythos AI model. These events collectively illustrate the varied nature of current cyber threats.

How are the Microsoft Defender zero-days being actively exploited?

Microsoft has rolled out security patches for two Microsoft Defender vulnerabilities, CVE-2026-41091 and CVE-2026-45498, that are currently under active exploitation as zero-days. The first flaw, CVE-2026-41091, is a privilege escalation vulnerability found in Microsoft Malware Protection Engine versions 1.1.26030.3008 and earlier. This issue stems from an improper link resolution before file access, also known as a link-following weakness, which allows threat actors to gain SYSTEM privileges on an affected system.

The second vulnerability, CVE-2026-45498, affects systems running Microsoft Defender Antimalware Platform versions 4.18.26030.3011 and earlier. Successful exploitation of this flaw enables attackers to trigger denial-of-service (DoS) states on unpatched Windows devices. Both of these components are integral to Microsoft's antivirus and antispyware software, including System Center Endpoint Protection and Security Essentials. These incidents follow previous disclosures of similar vulnerabilities, as detailed in analyses of Microsoft Defender zero-days and specific CVE-2026-33825 exploitation.

Microsoft has addressed these security flaws by releasing Malware Protection Engine versions 1.1.26040.8 and Antimalware Platform versions 4.18.26040.7, respectively. The default configuration of Microsoft antimalware software is designed to automatically keep malware definitions and the Windows Defender Antimalware Platform up to date, minimizing the need for manual intervention from most users. However, administrators should verify that automatic updates are configured correctly and confirm the installation of the latest versions through the Windows Security program. Further information on related incidents and vulnerabilities can be found in our previous discussion on CVE-2026-33825 and other actively exploited Windows zero-days.

The U.S. Cybersecurity and Infrastructure Security Agency (CISA) has added both CVE-2026-41091 and CVE-2026-45498 to its Known Exploited Vulnerabilities (KEV) Catalog. CISA has mandated that Federal Civilian Executive Branch (FCEB) agencies secure their Windows endpoints and servers against these zero-day vulnerabilities within two weeks, setting a deadline of June 3, 2026, as per Binding Operational Directive (BOD) 22-01. The agency emphasized that such vulnerabilities are frequent attack vectors for malicious cyber actors and pose significant risks to federal enterprises.

TeamPCP Breached GitHub Internal Repositories via VS Code Extension

GitHub has officially confirmed a breach of its internal repositories, attributing the compromise to a supply chain attack involving a malicious version of the Nx Console Microsoft Visual Studio Code (VS Code) extension. The cybercriminal group TeamPCP is responsible for exfiltrating approximately 3,800 repositories in this incident. This attack is part of a broader campaign that also affected other prominent technology companies, including OpenAI, Mistral AI, Grafana Labs, and other companies, stemming from the recent TanStack supply chain compromise.

The malicious extension, identified as nrwl.angular-console, was briefly available on the Visual Studio Marketplace for a mere 18 minutes, from 12:30 p.m. to 12:48 p.m. UTC on May 18, 2026. Despite this short window, the attackers successfully distributed a credential stealer. This malware was capable of harvesting sensitive data from various developer tools and services, including 1Password vaults, Anthropic Claude Code configurations, npm, GitHub, and Amazon Web Services (AWS) credentials.

The attack was executed by a hidden shell command disguised as a routine setup task, which downloaded and executed a hidden package from a planted commit on the official nrwl/nx GitHub repository. GitHub stated that there is no evidence of impact to customer information stored outside of its internal repositories, such as customer-owned enterprises, organizations, or repositories. However, internal repositories contained information from customers, including excerpts of support interactions, and GitHub confirmed it would notify affected customers if any impact is discovered.

TeamPCP has rapidly gained notoriety for its large-scale software supply chain attacks, specifically targeting widely used open-source projects and security-adjacent tools critical to developers. The group uses a self-sustaining cycle of compromises: breaching one trusted tool, stealing credentials from developer systems that install it, and then using those credentials to breach the next legitimate tool. The prevalence of auto-update mechanisms in popular extension marketplaces, like VS Code, facilitates rapid propagation of compromised software once a publisher is breached.

Microsoft Disrupts Fox Tempest Malware-Signing-as-a-Service Operation

Microsoft has successfully disrupted a malware-signing-as-a-service (MSaaS) operation attributed to the threat actor Fox Tempest. Codified as "OpFauxSign," this significant enforcement action targeted Fox Tempest's website, signspace[.]cloud, and involved taking offline hundreds of virtual machines crucial to the operation. This service enabled cybercriminals to disguise malware as legitimate software using fraudulently obtained code-signing certificates, thereby bypassing security controls. Fox Tempest has been active since May 2025.

The MSaaS operation weaponized Microsoft's Artifact Signing system, formerly known as Azure Trusted Signing, which is designed to ensure software legitimacy. Fox Tempest exploited this system to generate short-lived, fraudulent code-signing certificates, each valid for only 72 hours. These certificates were then used to sign malicious code, making it appear trusted. The threat actor likely used stolen identities based in the United States and Canada to obtain the necessary digital credentials for signing.

The service, which cost between $5,000 and $9,000, allowed paying cybercriminal customers to upload malicious files for signing. This facilitated the deployment of various malware families and ransomware. Ransomware groups and malware associated with Fox Tempest include:

• Rhysida ransomware, deployed by threat actors such as Vanilla Tempest.
• Other malware families: Oyster (aka Broomstick or CleanUpLoader), Lumma Stealer, Vidar, and related families.
• Ransomware strains with established connections: INC, Qilin, BlackByte, and Akira.

These operations have targeted critical sectors, including healthcare, education, government, and financial services, across multiple countries such as the U.S., France, India, and China. The signed malware often masqueraded as legitimate applications like AnyDesk, Microsoft Teams, PuTTY, and Cisco Webex. From February 2026, Fox Tempest shifted to providing pre-configured virtual machines (VMs) hosted on Cloudzy, simplifying the delivery of malicious binaries.

What information did ShinyHunters steal from 7-Eleven?

The convenience store giant 7-Eleven confirmed a data breach following claims by the notorious cybercriminal organization ShinyHunters. The breach specifically targeted certain 7-Eleven systems utilized for storing documents related to its franchise operations. The company disclosed the incident to state regulators in Maine, Vermont, and Massachusetts this week, although the total number of affected individuals has not been publicly specified.

The information exfiltrated by ShinyHunters includes sensitive Personally Identifiable Information (PII) belonging to franchisees. The stolen data comprises:

• Names
• Addresses
• Social Security numbers
• Other PII

7-Eleven, which operates approximately 86,000 stores across 19 countries, with nearly 10,000 franchises in the U.S., discovered the breach on April 8, 2026. Subsequent investigation confirmed that cybercriminals had gained unauthorized access to these systems. The breach notification letters confirm the type of data compromised.

The confirmation of this breach comes after ShinyHunters claimed responsibility for stealing troves of information from 7-Eleven on Salesforce platforms in late April. ShinyHunters is a cybercriminal group specializing in large-scale data breaches and extortion, frequently targeting data storage tools of high-profile companies across technology, finance, retail, and other sectors. The FBI recently released a public service announcement (PSA) urging victims of ShinyHunters to refrain from paying ransoms, showing the group's tendency to sell stolen data or reuse it for further attacks.

U.S. Regulators Halt Bank Cybersecurity Exams Due to Mythos AI Risks

U.S. financial regulators have paused certain cyber-related examinations of the nation's largest banks due to emerging risks posed by Anthropic PBC's new Mythos AI model. The Federal Reserve and the Office of the Comptroller of the Currency (OCC) have temporarily suspended some of these examinations to provide financial institutions, including JPMorgan Chase & Co., Morgan Stanley, Goldman Sachs Group Inc., and other large banks, with more time to assess and bolster their systems against the new threats.

The concern stems from Mythos AI's reported capability to rapidly move through code and identify hacking weaknesses, a power that alarmed some bank executives. Anthropic PBC had previously limited access to Mythos, warning of its potential use in cyberattacks, and initiated "Project Glasswing" with a select few companies, including Apple Inc. and JPMorgan Chase & Co., to test their cyber defenses. This initiative allows these firms to gain early access to the technology for defensive purposes.

Treasury Secretary Scott Bessent and then-Fed Chair Jerome Powell had previously convened Wall Street leaders to discuss the cyber risks introduced by AI models like Mythos. Following these discussions, banks and regulatory officials have been collaborating to methodically address the vulnerabilities identified. The pause in examinations is intended to allow banks to fully comprehend the implications of this new technology and enable regulators, such as the OCC which is conducting its own trial run with Mythos, to stress-test existing defenses.

Leading U.S. banks with access to Mythos have established specialized, secretive teams dedicated to working with the AI model. Many are also engaging directly with federal intelligence agencies to map out potential threats. This collaborative effort extends to security vendors, as mentioned by Goldman Chief Executive Officer David Solomon and JPMorgan CEO Jamie Dimon, who stated that hundreds of people are engaged full-time on this work.

Technical Takeaways

• Immediate patching is needed for zero-day vulnerabilities in core security components like Microsoft Defender, particularly those enabling privilege escalation and denial-of-service, across all affected Windows environments, especially given active exploitation.
• Supply chain attacks using developer tools, such as the Nx Console VS Code extension in the GitHub breach, remain a critical way for sophisticated threat actors like TeamPCP to exfiltrate sensitive data and credentials from multiple high-profile targets.
• Disrupting Malware-Signing-as-a-Service (MSaaS) operations, like Fox Tempest's, directly affects the ability of numerous ransomware groups and malware families to distribute trusted, signed malicious code. This shows the importance of targeting cybercrime infrastructure.
• The proactive pausing of cybersecurity examinations for major financial institutions due to advanced AI models like Mythos AI shows a new approach to risk assessment. Here, theoretical AI capabilities can prompt significant shifts in regulatory and corporate security postures.
• Data breaches by financially motivated groups such as ShinyHunters continue to target Personally Identifiable Information (PII), demonstrating the persistent threat to sensitive customer and franchise data held by organizations.