Weekly Threat Digest 26.47 || Nov 16
ScreenConnect hijacks hit MSPs, while Outlook and Chrome flaws join the KEV catalog
Attackers hijacked cloud-hosted ConnectWise ScreenConnect panels to push rogue clients and the SCKit toolkit to managed endpoints, affecting at least 288 hosts across 48 organizations, according to Huntress. Microsoft's Patch Tuesday fixed an exploited Outlook spoofing flaw, CVE-2026-124194, and Google fixed an exploited Chrome flaw, CVE-2026-125350. CISA added both to its KEV catalog on November 10.
Siemens published advisories for energy management, grid, and building automation software affected by the PHP2Shell and Node2Shell flaws covered in earlier issues. N-able took its hosted N-central service offline after a researcher published 33 critical flaws, Fortinet patched dozens of flaws with no known exploitation, and Node.js fixed 16 flaws across its supported release lines. Elsewhere, Unit 42 reports government webmail compromises through injected JavaScript, and SafeDep tracks a credential stealer pushed into more than 8,000 GitHub repositories.
Act Now
- ConnectWise ScreenConnect: Attackers hijacked MSP instances to push rogue clients and the SCKit toolkit to at least 288 hosts. If you run ScreenConnect: keep RMM+ extensions disabled, update technician clients to 26.6.5 or later, and investigate suspect hosts out-of-band.
- Google Chrome: CVE-2026-125350, a use-after-free flaw in Chrome's XML component, is exploited in the wild and was added to KEV on November 10. If you run Chrome: update to 157.0.8103.50 or later, prioritizing machines that browse the open web.
- Microsoft Outlook and Office: An exploited Outlook spoofing flaw (CVE-2026-124194, CVSS v3.1 6.5) was added to KEV on November 10. If you run Office LTSC 2024 or 2021: update to 16.0.17932.21140 or 16.0.14334.21048 or later, respectively; for Microsoft 365 Apps, apply the current security update.
Drafted with AI from public reporting; reviewed and edited by GPP Labs before publishing.
OT/ICS Advisories
Marker shows the highest CVSS v3 score in each advisory: 🔴 9.0 and up · 🟠 7.0–8.9 · 🟡 4.0–6.9
- 🔴 Siemens Spectrum Power and Energy Engage (energy management and utility software) · ICSA-26-316-01 · 2 CVEs · CVSS v3.1 9.8. Used in energy sector control and monitoring. The PHP2Shell flaws in the PHP runtime allow unauthenticated remote code execution, and both are on CISA's KEV catalog. Upgrade PHP to 8.5.12 or apply Red Hat's patch; Siemens has no fix planned for some products CISA
- 🔴 Siemens RUGGEDCOM APE1808 (industrial application hosting platform) with Palo Alto Networks PAN-OS · ICSA-26-316-05 and ICSA-26-316-04 · 3 CVEs · CVSS v3.1 9.8. Two PHP2Shell flaws in PAN-OS allow remote code execution, and both are on CISA's KEV catalog; a separate advisory covers the Brotli flaw known as Node2Shell (CVE-2026-111497). Contact Siemens customer support for the applicable PAN-OS update CISA
- 🔴 Siemens Desigo CC (building management), Gridscale X (grid planning and operations software), and Industrial Edge Management · ICSA-26-316-02 and ICSA-26-316-03 · 1 CVE · CVSS v3.1 9.8. The Brotli decoder flaw known as Node2Shell (CVE-2026-111497), first covered in 26.44, is an out-of-bounds write that lets a remote attacker corrupt memory and potentially run arbitrary code. Update Gridscale X PSS®E to V36.8.0 or later and DSA to V5.4 or later, and install the latest .NET security updates for Desigo CC. For Industrial Edge Management, Siemens has fixed several configurators (such as Databus Configurator 4.0.1) and recommends mitigations where Management or Device Applications run without a fix; setups with no Management Applications need no action CISA
Vulnerabilities & Exploits
Attackers hijack MSPs' ScreenConnect instances to push rogue clients and SCKit
Huntress reports that on November 11, an operator used access to cloud-hosted ScreenConnect panels to push commands to managed endpoints, installing attacker-controlled ScreenConnect clients and a toolkit called SCKit. At least 288 hosts across 48 Huntress-protected organizations were affected. Microsoft Threat Intelligence also reported the activity, and ConnectWise says the initial access route is still under investigation.
Huntress says the activity appears related to the RMM+ extension, whose versions 1.0.31 and 1.0.32 accepted a blank access key when none had been configured. SCKit replaces the real client and goes after the technicians who connect to infected hosts, with components built to steal their session cookies and spread through the console; Huntress has not seen that spread succeed.
- Exposure: Cloud-hosted and on-premises ScreenConnect instances running the RMM+ extension family (Labtech Plus, RMM Plus, RMM+, and the two Password Link extensions), plus technician workstations running host clients older than 26.6.5, which are exposed to CVE-2026-84869 (on CISA's KEV catalog since September 11).
- Response: Keep those extensions disabled, update technician clients to 26.6.5 or later (a cloud console update does not update them), and investigate suspected hosts out-of-band rather than through ScreenConnect. If a technician connected to a suspected host, treat that workstation as exposed, revoke web sessions, and rotate credentials from a clean device.
- Detections: Microsoft published Defender hunting queries and IOCs, and Huntress published IOCs including file hashes and attacker infrastructure.
- Also reported by: Beazley Security
Microsoft patches exploited Outlook flaw; Windows Search bug rated 9.8
Microsoft's November Patch Tuesday fixed CVE-2026-124194, an Outlook spoofing flaw (CVSS v3.1 6.5, Moderate) that Microsoft says is exploited. CISA added it to its KEV catalog on November 10. Microsoft's advisory gives no details on how attackers use it or whom they target.
The SANS Internet Storm Center counts 1,260 fixes this month, 68 of them critical, including 167 Chromium issues. Critical flaws include remote code execution in Windows NAT, Active Directory Domain Services, Exchange Server, and Schannel. ZDI's Dustin Childs told Dark Reading that a Windows Search remote code execution flaw (CVE-2026-116928, CVSS 9.8), reachable through Exchange Server, needs immediate attention.
- Exposure: Microsoft Office LTSC 2024 before 16.0.17932.21140, Office LTSC 2021 before 16.0.14334.21048, and Microsoft 365 Apps for Enterprise, plus Windows and Exchange Server systems missing the November updates.
- Response: Update Office LTSC 2024 to 16.0.17932.21140 or later and Office LTSC 2021 to 16.0.14334.21048 or later, and apply the current Microsoft 365 Apps security update. Install the November Windows updates, and update Exchange Server Subscription Edition to 15.02.2562.057 or later.
- Also reported by: Center for Internet Security
Fortinet patches dozens of flaws across FortiMail, FortiOS, FortiManager
Fortinet published advisories on November 10 for dozens of vulnerabilities across FortiMail, FortiOS, FortiManager, FortiAnalyzer, and FortiAuthenticator. The worst are two unauthenticated FortiMail flaws rated CVSS v3.1 9.8: a path traversal and a command injection in HA join requests. Fortinet says none is known to be exploited, and it credits most of the findings to internal discovery using the AI model Claude Mythos.
Other critical flaws include FortiMail SQL injection and account takeover, a FortiManager code execution bug (CVSS v3.1 9.0), and a FortiAuthenticator hard-coded password bypass. MS-ISAC also reports no exploitation in the wild.
- Exposure: Affected versions vary by advisory; FortiMail 7.2 has no fixed release.
- Response: Upgrade to the fixed release in each Fortinet advisory, and migrate off FortiMail 7.2. Where the JSON API isn't needed on FortiManager or FortiAnalyzer, Fortinet says setting
rpc-permitto none on admin profiles reduces exposure.
N-able N-central "Able2Shell" flaws force cloud offline
Huntress reports that N-able took its hosted N-central environment offline on November 9 after an independent researcher published 33 critical vulnerabilities in the remote monitoring and management (RMM) platform to GitHub. N-able said it had seen no active exploitation as of November 9 at 15:15 UTC.
The flaws can be chained into four exploits dubbed "Able2Shell": three lead to remote code execution and one to privilege escalation to root. They affect N-central Appliance 2026.4.0.31 and likely earlier versions. N-able says a hotfix is coming soon but has not given a date. Three earlier N-central flaws this year (CVE-2026-18556, CVE-2026-18577, and CVE-2026-86218) are on CISA's KEV catalog.
- Exposure: N-central Appliance 2026.4.0.31 or earlier on premises, and the hosted service, now offline; managed service providers' downstream clients may also be affected.
- Response: If you run N-central on premises, take the server offline and keep it offline until N-able confirms it is safe, then apply N-able's hotfix (no fixed version has been named). Review login and authentication activity from November 8 onward, and watch for anomalous remote access sessions, newly deployed scripts, and unauthorized administrator accounts.
More This Week
- Chrome fixes an exploited XML flaw now on KEV (CVE-2026-125350): Chrome 157.0.8103.50 fixes nine flaws, including a use-after-free in Chrome's XML component that Google says is exploited in the wild. CISA added it to KEV on November 10. Update to 157.0.8103.50 or later (157.0.8103.50/.51 on Windows and Mac). Center for Internet Security
- Node.js fixes 16 flaws across its supported release lines (16 CVEs): Node.js 22.23.5, 24.22.3, and 26.13.1, released November 11, fix seven High, six Medium, and three Low issues, mostly in HTTP/2, HTTPS, and TLS. One High issue, CVE-2026-123864, lets an unauthenticated client read memory belonging to other connections, including TLS private keys, from HTTPS servers whose handlers flush headers early. Upgrade and restart Node.js processes; end-of-life lines get no fix. Node.js
- SonicWall patches seven SonicOS flaws, including a KEV-listed SSLVPN bypass (7 CVEs): An MS-ISAC advisory covers seven SonicOS flaws in Gen7 through 7.3.3-7015 and Gen8 through 8.2.2-8015, among them the SSLVPN authentication bypass (CVE-2026-19868) whose KEV listing 26.46 covered. Apply SonicWall's fixed versions. Center for Internet Security
- Apache HTTP Server mod_http2 race condition allows remote code execution (CVE-2026-119618): Servers with HTTP/2 enabled are exposed to unauthenticated remote code execution. Upgrade to Apache HTTP Server 2.4.71 or later, or set Protocols to http/1.1 until patched. No exploitation is known. Fortinet
- GPUStack authentication bypass added to KEV as exploit attempts begin (CVE-2026-102895): A spoofed X-Gpustack-Real-IP header lets attackers bypass authentication on the GPU cluster manager. CISA added the flaw to KEV this week, and exploit attempts are underway. Update to GPUStack 2.2.0 or later. SANS Internet Storm Center
- Also patched this week, with no exploitation reported: Apple fixed nine flaws, several of them Bluetooth issues that may weaken device pairing, in iOS and iPadOS 27.2.1 (26.7.3 or 18.7.12 on older devices) and macOS Golden Gate 27.2.1, Tahoe 26.7.3, and Sequoia 15.8.3 SANS Internet Storm Center. Mozilla fixed code execution flaws in Firefox 160, Firefox ESR 153.7, and ESR 115.45 CIS: Mozilla. Adobe fixed code execution flaws in Connect, Experience Manager, Photoshop, After Effects, Premiere, and other products CIS: Adobe. SAP's November updates fix flaws across NetWeaver, Web Dispatcher, Gateway, and other products, the worst allowing code execution CIS: SAP.
- Also added to CISA's KEV catalog this week: Check Point Multiple Products (CVE-2026-123039, Nov 13), Barracuda Email Security Gateway (CVE-2026-116606, Nov 10), Barracuda Networks Multiple Products (CVE-2026-111560, Nov 10), Cisco BroadWorks (CVE-2026-20320, Nov 10), IBM Aspera Faspex (CVE-2026-123218, Nov 9), SignalWire FreeSwitch (CVE-2026-116779, Nov 9), Openwave Webtop Media Transcoder (CVE-2026-111793, Nov 10), and UnrealIRCd (CVE-2026-114713, Nov 13). Federal deadlines run from November 12 to 24. CISA
Updates on Earlier Stories
ARTEX AI attacks turn exposed APIs into cloud access
Covered in 26.43, 26.44, and 26.46.
Unit 42 reports that attackers are using the ARTEX agentic AI framework to turn public password reset pages and exposed APIs into cloud access, sometimes in under a day, without exploiting a software vulnerability. It saw the activity across multiple organizations, with its threat hunting team tracking a cloud campaign against retail and consulting firms, and counted 609 internet-facing hosts running ARTEX as of October 2026.
Unit 42 describes the marks of an AI-driven operator: bursts of requests followed by 25- to 31-second pauses, pivots within about a minute of a credential download, and more than 200 client secrets validated in about three minutes. Because ARTEX is open source and widely forked, Unit 42 says indicator-based detection will not keep up.
- Exposure: Internet-facing applications that publish OpenAPI specifications, source maps, or configuration files with tokens, or that have unauthenticated write endpoints, weak self-service password reset, or long-lived cloud keys.
- Response: Remove public specifications and source maps, require authentication on every write endpoint (including development and staging), and rate-limit password reset and verification code endpoints. Replace long-lived cloud keys with role-based access, and hunt for test usernames in identity provider logs and first use of a cloud key from a new IP address.
- Also reported by: Cryptika
More Updates
- OpenSSH "sshhole" flaw added to KEV (CVE-2026-123158): CISA added the flaw to its KEV catalog on November 9, after 26.46 went out. SANS honeypots logged their first exploit attempt on November 5, and login attempts with overlong passwords peaked at 713 on November 6. Upgrade to OpenSSH 10.6p2 or turn off password authentication on affected musl-based hosts. First covered in 26.46. SANS Internet Storm Center
- WordPress Core flaw draws 240,000 exploit attempts (CVE-2026-119977): Security Boulevard reports 240,000 exploit attempts against this race condition, which allows remote code execution during core upgrades, and Fortinet says it is monitoring exploitation attempts. Upgrade to WordPress 7.1.4 and block web access to wp-content/upgrade/ before updating. First covered in 26.46. Fortinet · Security Boulevard
- Unpatched Danfoss refrigeration controllers exploited in cold-chain attacks: Sekoia ties the attacks to exposed, outdated Danfoss controllers but could not confirm whether the "code of the day" flaw (CVE-2025-41450) was used. It notes that French authorities attributed the attacks to an Iran-linked Islamic Revolutionary Guard Corps unit. Apply Danfoss firmware updates and remove controllers from direct internet exposure. Covered in 26.43, 26.44, and 26.46. Sekoia
Credential Theft & Phishing
BluePhishing abuses Entra Agent ID consent for initial access
Proofpoint researchers describe "BluePhishing," a proof-of-concept technique in which attackers trick a victim into consenting to a malicious Microsoft Entra agent blueprint. Consent creates a blueprint principal in the victim's tenant, which can then create agent identities and obtain tokens. Proofpoint has not seen in-the-wild use and notified Microsoft before publishing.
The blueprint principal automatically receives the AgentIdentity.CreateAsManager permission, which cannot be revoked. An attacker needs only the victim's tenant ID and a sponsor user, both discoverable without authentication. One blueprint can spawn up to 250 agent identities, which can inherit its permissions, and inherited permissions do not appear in the Entra portal or Microsoft Graph. Microsoft message MC1506733 says default users will be able to create blueprints starting in November 2026.
- Exposure: Microsoft Entra tenants that allow users to consent to third-party apps, which extends to agent blueprints and their agent identities. Admin consent, which requires a user with admin consent rights, gives the broadest access.
- Response: Restrict user consent to third-party apps, use permission classifications to limit what users can approve, apply Conditional Access and Entra ID Protection policies for agents, and remove known malicious agents by blueprint ID.
- Detections: Proofpoint documented audit log events to watch: "Add application" for blueprint creation and "Add service principal" for agent identity creation.
ClickFix attacks now hide malware in PNG images
Huntress reports that ClickFix campaigns, which trick users into pasting commands copied from a fake web verification page, now rebuild full executables from data hidden in PNG image pixels. In other cases the commands led to attackers moving from a user's PC to domain controllers. Huntress says the activity does not appear to be one group's work, and it names no attribution.
In the PNG chain, a PowerShell stager dropped a signed Microsoft binary (LockScreenContentServer.exe) beside a rogue DLL, then fetched the images and rebuilt the payloads. In a separate September intrusion, the pasted command ran Active Directory reconnaissance and set up reverse SSH tunnels; the attacker then used WinRM (Windows Remote Management) to reach the primary domain controller and created a new user account. Another campaign delivered DuskDesk, a Node.js remote access trojan (RAT) disguised as a remote management tool.
- Exposure: Windows environments where users can paste commands into the Run dialog or a terminal, especially where domain controllers are reachable from user endpoints.
- Response: If you can, restrict execution from user-writable locations such as Temp with application control, and alert on browsers or user sessions spawning
cmd.exeor PowerShell, new scheduled tasks or Run keys, and unexpected SSH connections. Train users never to paste commands at a web page's request. - Detections: Huntress published IOCs for all three attack chains, including hashes, domains, IP addresses, and persistence names.
- Also reported by: Cryptika
Quick Hits
- Fake Microsoft Teams booking flow delivers device code phishing: A spear-phishing lure disguises a Microsoft device-code request as a Teams meeting booking, tricking users into authorizing the attacker's session. Restrict device-code sign-in with Conditional Access, enforce phishing-resistant MFA, and investigate unfamiliar device sign-ins. Huntress
- Macfinger ClickFix campaign delivers DownTheDRAIN infostealer to macOS: A fake CAPTCHA page tricks users into pasting a Terminal command that installs the infostealer. The SANS Internet Storm Center analyst's macOS 26.7.3 system moved the malware to the Trash before it could exfiltrate sensitive data. SANS Internet Storm Center
Ransomware
Citrus ransomware enters through contractor accounts and hides in VLESS tunnels
Kaspersky's Securelist reports a previously undocumented ransomware group, Citrus, that gets into networks with credentials stolen from contractors' employees, connecting through corporate VPNs or internet-facing RDP servers. Kaspersky says it attacks only Russian companies, but the tradecraft travels. To escalate from limited contractor accounts, the group tried to exploit an AD CS authentication bypass (CVE-2026-54121, "Certighost"), which leaves machine accounts named GHOST plus eight random letters. It tunnels out with Xray reverse VLESS connections on port 443, disguised as traffic to Yandex sites, and steals Telegram session data before encrypting with the Go-based Sn00py ransomware. A malicious Group Policy Object named "Kaspersky GPO" spreads the ransomware and turns off Windows Defender. Kaspersky infers from script comments that LLMs helped write commands and scripts, and it notes overlaps with the Toy Ghouls group.
- Exposure: Organizations that give contractors VPN or RDP access, especially domains running Active Directory Certificate Services (AD CS).
- Response: Restrict and monitor contractor access, and alert on new machine accounts named GHOST followed by random letters, unexpected Group Policy Objects or executables in SYSVOL, Xray or VLESS tunnels, and policy changes that turn off Defender.
- Detections: Kaspersky published file hashes, file paths, scheduled task names, and IP addresses.
Ransomware Tracker
Ransomware and data-extortion groups: group · sectors when named · what's new this week
- Qilin · Hexastrike Cybersecurity describes S5PIVOT, a Python backconnect proxy found hidden inside a Qilin ransomware intrusion. Hexastrike Cybersecurity
- PixelLeaks · CyberXTron describes growing leak activity from the PixelLeaks ransomware group. CyberXTron
Threat Activity & Tradecraft
wfx campaign plants GitHub Actions credential stealer in 8,000 repositories
SafeDep reports that an attacker used stolen GitHub tokens to push a credential stealer into 8,093 repositories between November 10 and 12. The payload runs in GitHub Actions or on npm install, sends secrets and cloud credentials to the attacker, and tries to leave AWS and Azure backdoors that survive secret rotation. Aikido Security first reported the campaign, and SafeDep says it was still active at the time of its analysis.
In almost all repositories the attacker then reverts the commit, so the branch looks clean. On Azure, the payload adds a federated credential named wfx-ci-<n> to Entra app registrations, which lets any workflow on the repository's main branch sign in as the app even after client secrets are rotated. Two packages carry the payload: digitalelvis/neuronai-studio v3.14.3 on Packagist and the latest github[.]com/vkuznet/gocurl on the Go module proxy.
- Exposure: Public GitHub repositories that received commits such as
ci: refresh build cacheorchore: add ci helper, awfx-docsbranch, or a pull request titleddocs: dependency audit notes, plus anyone who rannpm installon those commits. - Response: If a repository is affected, revoke the commit author's tokens and sessions, rotate every secret the repository can read, and check Entra ID for
wfx-ci-*federated credentials you did not create. Pull any package version built between the malicious commit and theci: cleanupcommit, and republish from a clean commit. - Detections: SafeDep published 38 indicators of compromise, a searchable list of affected repositories, and audit log checks for Entra ID, AWS, GCP, and Vault.
- Also reported by: Aikido, Step Security
Falcon C2 framework hijacks government webmail with injected JavaScript
Unit 42 reports that an actor used Falcon (also called XG or Xiàng Gē), a browser-based command and control (C2) framework, to compromise government webmail portals and web applications by planting malicious JavaScript in the sites. Between March and May 2026, it found compromises at government entities in one Middle Eastern country and four Southeast Asian countries, including ministries of foreign affairs, defense, and interior, and law enforcement agencies. Unit 42 assesses with high confidence that Falcon and the earlier PeckBirdy framework are variants of the same core framework.
In the Middle Eastern country, a script tag injected into a foreign ministry's Roundcube webmail portal loaded the implant and captured credentials, and 39 Roundcube instances there were affected. Elsewhere, a sideloaded Chrome extension posing as "PDF Viewer" took credentials and screenshots, and a fake Adobe Flash update delivered Antino, a Rust implant that uses Microsoft OneDrive for C2.
- Exposure: Government agencies and other organizations running Roundcube or other internal web portals, plus browsers with developer-mode extensions that mimic utilities and request broad permissions.
- Response: Check internal portal templates and headers for script tags loading unfamiliar domains, audit browsers for developer-mode extensions posing as PDF viewers, and look for
Hm_lvt_cookies or local-storage keys on internal sites and OAuth activity to OneDrive tenants outside your own. - Detections: Unit 42 published IOCs, including C2 and payload domains, a Chrome extension ID, and an Antino sample hash.
Worth Watching
- Attackers exploit AhsayCBS flaws to plant webshells, a miner, and a C2 implant: Huntress details an intrusion through two KEV-listed AhsayCBS flaws (CVE-2026-105133 and CVE-2026-105134), with five webshell families, two miners, and a Python implant. Apply hotfix 10.3.4.45 (v10) or 9.15.2.70 (v7 to v9), then hunt for planted files. Huntress · Cryptika
- Attackers hunt for SAP NetWeaver webshells left by CVE-2025-31324: SANS honeypots show attackers trying to exploit the KEV-listed SAP NetWeaver file upload flaw, or to find webshells it left behind. Check
/irj/servlet_jsp/irj/root/for unexpected files. SANS Internet Storm Center - T-SQL abused to deliver miner and remote control on compromised SQL Servers: Fortinet describes a campaign that brute-forces internet-facing SQL Servers on port 1433, enables xp_cmdshell, and drops a Monero miner and a Gh0st RAT variant. Fortinet
- Threat actors use GSocket for persistent C2 on compromised Linux servers: Microsoft says attackers deploy GSocket after exploiting web applications, so access survives patching of the original flaw. Traffic uses a public relay, leaving no attacker-owned address to block. Remove the client and its persistence. Microsoft Threat Intelligence
- Attackers use AI to revive stalled intrusion on a U.S. retailer: Symantec assesses with high confidence that attackers used AI-generated scripts to speed credential theft and lateral movement after hand-typed commands stalled. It assesses with moderate confidence that an operator stayed in the loop. Symantec
- Chinese-speaking attackers use autonomous AI framework against semiconductor, defense, and civil society targets: Unit 42 reports that attackers ran the open-source Cairn framework with Chinese AI models in 25 operations against at least 19 organizations, gaining access at 10. It assesses with moderate confidence this is China-nexus intelligence collection. Unit 42
Every item links to its original source. Spot an error? Reply or leave a comment. Know a defender who'd find this useful? Forward it along.
GPP Labs · Threat Intelligence & Detection Engineering
GPP // Genuine People Personality™