Curated cybersecurity briefings on active threats, vulnerabilities, and emerging risks
Daily Cybersecurity Briefing (21 November 2026)
At a Glance
- Broadcom patched VMware vCenter flaw CVE-2026-123890, an authentication bypass exploited as a zero-day by China-linked Armory Castle since early September.
- JPMorgan, Visa and Mastercard attributed exposed customer marketing data to third-party providers after the N0n extortion group leaked allegedly stolen files.
- The European Commission confirmed an AI-assisted cyberattack disrupted its services, forcing several applications offline, though it has found no evidence of a data breach.
- A China-linked actor used the open-source Cairn AI framework to find and exploit a MONAI Label zero-day in university attacks.
- Five nginx memory corruption flaws, collectively named nginx2root, let unauthenticated attackers crash workers or run commands on affected servers.
- N-able confirmed unauthorised access to a Mail Assure and SpamExperts Hosted Cloud server caused a global outage, with dashboard and API access still offline.
Editorial Analysis
The vCenter zero-day, CVE-2026-123890, was exploited for more than two months before a fix became available. Google Threat Intelligence Group attributes the activity to Armory Castle, formerly UNC5221, a suspected China-nexus group that has used the flaw since at least early September. The authentication bypass affects VMware Directory Service and carries a CVSS score of 9.8. It allows a remote, unauthenticated attacker to obtain an administrator session on vCenter Server, the platform used to manage fleets of virtual machines. Broadcom's advisory VMSA-2026-0009 also covers CVE-2026-123891, a separate vulnerability that allows attackers to read and write the SDDC Manager database, execute commands and escalate privileges to root. This second flaw has not been linked to the reported exploitation. The gap of more than two months between the start of exploitation and the availability of a patch is the central concern. The sources do not specify how many environments were affected.
The N0n leak and N-able incident both involve third parties holding data or access, while the affected companies have disclosed limited information about the scope. JPMorgan, Visa and Mastercard each attribute the exposure to external service providers. None has named those providers or said how many customers were affected. JPMorgan says attackers accessed marketing data that may include names and email addresses. Reuters' review of files linked to JPMorgan found more than 16,000 customer email addresses, some accompanied by phone numbers and postcodes, as well as what appear to be 124 employee addresses. SOCRadar, which analysed all 14 files in the Visa and Mastercard archives, found no card numbers, expiry dates, passwords or home addresses. N0n's claim that the leak involved "hundreds of thousands" of campaign interactions comes from the group itself and remains unverified. N-able's timeline shows a different kind of revision. The company initially attributed a service disruption reported on 16 November to platform configuration and service dependencies. On 18 November, it said unauthorised access to a server was the cause. The dashboard, API, quarantine and logging remain offline, while mail flow is unaffected. N-able reports no evidence that customer data, including tenant mailboxes, was compromised.
The Cairn case describes autonomous AI agents identifying and exploiting a previously unknown vulnerability. According to Amazon CISO CJ Moses, a suspected China-linked actor used the open-source penetration-testing framework Cairn to attack universities in East Asia outside mainland China. The agents identified a zero-day in MONAI Label, an open-source medical imaging AI application, and used it to gain full control of a server. Amazon's Threat Intelligence Team discovered the campaign on 10 September through a publicly hosted Cairn instance that exposed its activity. The maintainers have since released a fix and an advisory. Some reports describe the vulnerability as an apparent or possible zero-day, so the extent of the agents' autonomy in discovering it remains an assessment by Amazon, not an independently verified finding. The European Commission incident is separate. The Commission says a publicly available AI tool was misused in the attack, taking several applications offline. It has found no evidence of a data breach but has not identified the tool involved.
Highlights of the Day
JPMorgan, Visa and Mastercard Cite Vendor Breaches After Extortion Claims
The N0n extortion group claimed to have stolen customer data tied to JPMorgan Chase, Visa and Mastercard, which each attributed the exposure to third-party service providers. JPMorgan said attackers accessed marketing data that may include customer names and email addresses, while Visa and Mastercard reported no impact to payment data or networks. N0n's leak site marks all three companies as non-paying and offers the allegedly stolen data for download.
WhatsApp Fixes Android SQL Injection Flaw Triggered by Crafted Messages
WhatsApp patched CVE-2026-102769, a SQL injection flaw in Android versions before 2.26.35.74, triggered by crafted order details messages from unofficial clients. The bug could delete locally stored files and leak one bit of chat database information, with no exploitation seen. The November update also fixes an attestation certificate flaw affecting Meta AI and an ad-reply bug exposing device IP information.
West Virginia Legislature Cyberattack Takes Website and IT Systems Offline
The West Virginia Legislature is responding to a cyberattack that began on Thursday and continued into Friday, taking its website offline and disrupting IT systems. Officials temporarily disabled access to certain systems and are investigating with outside cybersecurity specialists. They said the incident appears limited to Legislature systems, and no group has claimed responsibility.
N-able Confirms Security Incident Affecting Mail Assure and SpamExperts Hosted Cloud
N-able confirmed unauthorised access to one server in its Mail Assure and SpamExperts Hosted Cloud environments across all regions. The company initially attributed the disruption that began on 16 November to platform configuration and service dependencies, before identifying a security incident. The dashboard, API, quarantine and logging remain offline, while mail flow and filtering continue to operate, and N-able reports no evidence of customer data compromise.
Com-affiliated groups, including Bling Libra and newly tracked clusters Vortex and Cinder, pose as IT support to direct employees to adversary-in-the-middle phishing pages that capture credentials and MFA approvals. Attackers then enrol their own MFA devices, exfiltrate Microsoft 365 data and send extortion messages through Teams or email. Healthcare overtook financial services as the top target in September, while Australian organisations rose from under 1% to about 13% of targets in October.
Rent-a-Trojan Service Targets Indian Bank Customers With Fake Play Store Pages
A malware-as-a-service panel hosted 50 tenant accounts, 34 of them paid, letting fraudsters create fake Google Play pages and branded Android banking malware aimed at six Indian banks and card issuers. The lure pages collect card details, then push a dropper that disables Google Play Protect and installs a hidden SMS stealer. The stealer forwards OTPs to the attackers and sets call forwarding so bank verification calls reach them.
Daily Coverage