CVE-2019-18935, CVE-2026-13181, CVE-2026-13182, CVE-2026-67276, CVE-2026-67277, CVE-2026-67278, CVE-2026-67279, CVE-2026-67281, CVE-2026-75650, CVE-2026-82329, CVE-2026-83548, CVE-2026-83549, CVE-2026-85046, CVE-2026-86060, CVE-2026-86206, CVE-2026-86207, CVE-2026-86218
Domains:
assignpasskey[.]com, mfaregister[.]com, passkeydeploy[.]com, readthedocs[.]io, github[.]io, 30[.]232, xfjcc[.]fun
Get tomorrow's brief in your inbox
Today: Google patched a Chrome zero-day already under active exploit. Adobe Commerce sites face a maximum-severity zero-day deploying backdoors in production. N-able fixed critical N-central flaws after a customer compromise, and CISA added three SonicWall bugs to the KEV catalog with a September 18 deadline.
Chrome Zero-Day Under Active Exploitation (CVE-2026-85046)
Google released Chrome 152.0.7977.82 to patch 12 vulnerabilities, including a high-severity type confusion bug in V8 that has been exploited in the wild. The flaw allows remote code execution inside the sandbox via a crafted HTML page. This is the sixth actively exploited Chrome zero-day Google has addressed in 2026.
Adobe Commerce Zero-Day Deploying Backdoors (CVE-2026-75650)
Adobe patched a maximum-severity (CVSS 10.0) flaw in Adobe Commerce and Magento Open Source exploited in the wild starting September 4. The StyleSmuggler vulnerability abuses Magento's template system to inject PHP code via a "Payment Transaction Failed Reminder" email, achieving unauthenticated remote code execution. Sansec observed deployment of a Rust-based Linux backdoor and PHP web shells. Affected versions include 2.4.9-2026-aug and earlier across all product lines.
[kworker/u:8:0] or fc-cache and review authentication logs for anomalous admin access.SonicWall SMA Gateways on CISA KEV (CVE-2026-83548, CVE-2026-83549)
SonicWall addressed two critical vulnerabilities in SMA 1000 remote access gateways exploited as zero-days. CVE-2026-83548 is a pre-authentication SSRF rated CVSS 10.0, while CVE-2026-83549 enables post-authentication RCE. Both affect SMA 6210, 7210, and 8200v appliances and now appear on CISA KEV with a September 5 due date (already passed). EPSS scores are 0.007 (51st percentile) and 0.016 (75th percentile).
JFrog Artifactory Authentication Bypass Under Active Exploitation (CVE-2026-82329)
JFrog fixed a critical CVSS 9.8 authentication bypass in self-hosted Artifactory deployments that allows unauthenticated attackers to obtain administrator access tokens and control repositories. Exploitation began shortly after disclosure against internet-exposed systems. CISA added this to KEV with a September 5 deadline. EPSS score: 0.077 (94th percentile).
N-able N-central Customer Compromise
N-able released hotfixes for two severe N-central flaws (CVE-2026-86206, CVE-2026-86207) bypassing authentication controls and a maximum-severity RCE (CVE-2026-86218, CVSS 10.0) enabling pre-authenticated remote code execution. Huntress observed a fully patched N-central production environment compromised on September 4, likely via CVE-2026-86206 or CVE-2026-86207, though limited logging prevented definitive attribution. N-able states no confirmed production exploitation but warns unpatched systems remain at risk.
Microsoft 365 Vishing and AitM Token Theft (PREY-0058)
Arctic Wolf disclosed a widespread data theft and extortion cluster targeting Microsoft 365 through IT help desk vishing, adversary-in-the-middle token theft, and residential-proxy sign-ins. Attackers impersonate internal IT staff in phone calls, direct executives to authentication-themed domains (assignpasskey[.]com, mfaregister[.]com, passkeydeploy[.]com), and harvest credentials and MFA approvals. Stolen tokens are replayed from NodeMaven proxies matching victim geography. The threat actor performs bulk exfiltration from SharePoint, OneDrive, Exchange, and Box before sending extortion demands. Arctic Wolf links this activity to UNC6671 (Mandiant) and the Cinder/Pink extortion groups.
ScreenConnect Worm-Like VBScript Chain
Huntress disclosed worm-like activity abusing ConnectWise ScreenConnect to distribute a four-stage VBScript payload (1.vbs through 4.vbs) to newly connected systems. Three incidents used Quick Assist scams, phishing-delivered MSI installers, and fake Geek Squad refund forms for initial access. The VBScript chain profiles hosts, checks for security products (CrowdStrike, Huntress, SentinelOne, Sophos), downloads payloads from Dropbox based on system state, and deploys user-level ScreenConnect backdoors, UAC bypass tooling, or cryptocurrency miners. The malware spreads to newly connected hosts and terminates itself after final-stage execution to evade detection.
AI-Assisted Ransomware Intrusion in Under 10 Hours
Researchers detailed an AI-assisted ransomware intrusion that compromised an enterprise network in under 10 hours. Autonomous agents mapped internal systems, mined code repositories, obtained root credentials from a secrets manager, and abused build pipelines and cloud resources, compressing activity that normally requires substantially more human effort.
MikroTik RouterOS Zero-Days Exploited (MikroTrick Chain)
CERT Polska warned that attackers are actively exploiting two zero-day flaws in MikroTik RouterOS (CVE-2026-67276, CVE-2026-86060, CVSS 9.2) to bypass authentication and elevate privileges on devices supporting SSH remote access. The MikroTrick chain works without authentication when combined. Four additional CVEs were also patched (CVE-2026-67277, CVE-2026-67278, CVE-2026-67279, CVE-2026-67281). Fixed in versions 6.49.21 (Long-term), 7.23.4 (Long-term), and 7.24.2 (Stable).
Telerik UI Padding-Oracle Bug Chained to RCE
TantoSec published a working exploit chain targeting Telerik UI for ASP.NET AJAX (CVE-2026-13181, CVE-2026-13182, CVSS 8.1) enabling unauthenticated remote code execution. The chain abuses a padding-oracle vulnerability in the RadAsyncUpload control to forge encrypted upload configurations and load arbitrary DLLs. Exploitation requires a non-default configuration: a page rendering RadAsyncUpload whose handler reads upload results and an explicit encryption key (recommended by Telerik as hardening). Progress patched the flaws in version 2026.2.708 (July 8). No confirmed exploitation in the wild, but the release of tooling (telerik-rau-exploit) and payloads lowers the barrier for attackers. A 2019 deserialization flaw in the same handler (CVE-2019-18935) was previously exploited by ransomware operators and remains on CISA KEV.
BigBear 2.0 Phishing-as-a-Service Bypasses MFA at 258 Organizations
CloudSEK gained administrator access to the BigBear 2.0 phishing-as-a-service control panel and found the framework managed 42 VPS nodes targeting Microsoft 365, bypassing MFA at 258 organizations and stealing over 5,000 credentials. The service provides turnkey phishing infrastructure with multi-factor authentication bypass capabilities.
CrowdStrike Falcon Privilege Escalation on Windows 11 25H2 (FalconFlank)
Nightmare Eclipse disclosed FalconFlank, a zero-day privilege escalation affecting CrowdStrike Falcon on Windows 11 25H2 and Windows Server 2025. The proof-of-concept abuses Falcon's Microsoft Office macro-removal remediation behavior, allowing a low-privileged local user to obtain elevated access. CrowdStrike advised disabling the "Microsoft Office File Suspicious Macro Removal Windows policy setting" and stated customers remain protected through Cloud Anti-malware settings. A tech alert is available in the CrowdStrike support portal.
BengalSEO Malware Campaign Delivers MayaBot
The DFIR Report uncovered BengalSEO, an SEO poisoning campaign operating from Rajasthan, India since at least 2015. The operation uses black hat SEO techniques (backlinks, DOM injection, keyword stuffing) to promote lure pages impersonating technical support portals and software downloads on Microsoft Bing. Victims are routed through a traffic distribution system (TDS) using Matomo analytics for fingerprinting and delivered MayaBot malware (custom C2, system monitoring, XMRig miner) or tech support scams. Two IT service providers, WeConnect Solutions LLC and Garage2Global, develop the malicious web infrastructure. Lure domains include readthedocs[.]io sites and GitHub Pages (e.g., viziocomsetupentercode.github[.]io).
ClearFake WebDAV Infection Chain Delivers Amatera Stealer
Cisco Talos investigated a DLL named "verification.google" executing from WebDAV at a Ukrainian government organization, leading to discovery of a second loader ("pf.ch") and reconstruction of a ClearFake infection chain. The campaign uses Cloudflare Workers to inject JavaScript stored on BNB Smart Chain and a ClickFix prompt impersonating Google CAPTCHA, leading to Amatera stealer deployment. Secondary payloads include ZigCryptoStealer via NativeAOT loader and NetSupport Manager with Russian C2 infrastructure. Talos assesses with moderate confidence that the "verification.google" branch was conducted by a Russian threat actor (tracked as UAT-10820).
ClickFix Browser Cryptocurrency Theft via Google Sheets C2
Cisco Talos tracked a cryptocurrency-stealing campaign abusing the Google Visualization API for C2, retrieving obfuscated JavaScript from publicly published Google Sheets documents. The campaign uses ClickFix social engineering to convince targets to paste JavaScript into the Chrome address bar or install it via the Tampermonkey browser extension for persistence. The lure poses as a leaked vulnerability report for cryptocurrency swap services, targeting users willing to exploit it for financial gain. The injected script hooks the browser's fetch API, replaces cryptocurrency deposit addresses in server responses and the clipboard, and displays counterfeit "bonus" interface elements. Lures distributed via Telegram, DarkForums, and paste sites.
North Korean Linux Espionage Toolkit
Rapid7 disclosed a North Korean Linux toolkit targeting automotive and media organizations in South Korea. The framework includes a HAProxy instance (ted backdoor) and trojanized versions of agetty, atd, crond, polkitd, and sshd. Initial access via Groupware login portal exploitation, followed by SSH keylogger deployment for credential harvesting and lateral movement. The ted backdoor intercepts and injects HTTP traffic, executes C2 tasks, and achieves persistence. CurlRAT polls C2 every 12 hours for commands. Infrastructure mimics Naver's pstatic.net domain and overlaps with APT37 and Lazarus watering-hole techniques.
PEEP Post-Compromise Browser Backdoor
SOCRadar disclosed PEEP, a Chromium-based post-exploitation toolkit masquerading as a "Smart Bookmarks" extension. Requiring prior administrative access, the installer injects the extension into Chrome/Edge profiles by forging Secure Preferences integrity values. A native-messaging tool (nm_host.exe) extends functionality to host-level command execution and file management. The extension polls C2 (206.237.30[.]232 or xfjcc[.]fun) every 30 seconds over plaintext HTTP, exfiltrating browsing history, tab metadata, session cookies, and executing remote commands. PEEP is a derivative of the open-source RedExt framework with added installation routines, native host bridge, heartbeat telemetry, and broader command set. Chinese-language artifacts suggest a Chinese-speaking threat actor.
GitSpawn Vulnerability in AI Coding Agents
Researchers disclosed GitSpawn, a vulnerability class affecting AI coding agents including Claude Code, Codex, Cursor, Goose, Qwen Code, Grok Build, and Hermes. Malicious repository Git configurations can trigger arbitrary code execution as the developer when agents automatically gather project context, in some cases before trust prompts.
Avast, Nvidia Zero-Day Exploits Released
Nightmare Eclipse released PrettyPrague, a zero-day exploit targeting the Avast sandbox to spawn a shell with full system privileges. The exploit may also affect AVG and Norton (GenDigital products). GenDigital confirmed the vulnerability and released a fix, advising users to keep products updated. GreenSection exploits an out-of-bounds memory write in a shared global memory section used by Nvidia user-mode components. Nvidia is investigating through established security processes.
Thomson Reuters C-Track Breach Affects 11 US States, Canada
Thomson Reuters disclosed a breach of its C-Track court case-management platform affecting courts across 11 US states and Canada. An unauthorized party obtained C-Track files containing court records, including names and other personal information.
Baylor Genetics Data Breach Affects 2.8M Patients
Baylor Genetics disclosed a data breach affecting 2.8 million patients and employees after unauthorized access to part of its IT environment in June. Stolen data included names, birth dates, medical testing results, health insurance information, and some Social Security numbers.
Dropbox Account Access via Lenovo Email Verification Exploit
Dropbox disclosed unauthorized access to about 5,000 accounts after attackers exploited Lenovo's email verification process. Fraudulent Lenovo IDs created with victims' email addresses enabled access without Dropbox passwords, while files were viewed or downloaded from affected accounts.
Liquid Network $320M Bitcoin Theft
Hackers drained $320 million in Bitcoin from the Liquid Network sidechain. The attackers claim they are "the good guys," though the legitimacy of this claim remains unclear.
Cloud Misconfiguration Patterns Across AWS, Azure, Google Cloud
Intruder analyzed 3,000 organizations and found risk profiles differ dramatically by cloud provider. Weak IAM (80-98% of accounts) and missing logging (80-98%) are near-universal. AWS leads in exposed services (76%), permissive firewalls (83%), and weak encryption (49%). Azure leads in misconfigured services (80%) and storage account issues (61-67%). Google Cloud has the lowest prevalence across five of six categories. Top AWS issues: S3 not enforcing HTTPS (87%), permissive ACLs (84%), IAM privilege escalation paths (83%). Top Azure issues: Storage account key rotation disabled (67%), public network access enabled (61%), Entra users without MFA (55%).
Multiple zero-days under active exploitation this week (Chrome, Adobe Commerce, SonicWall SMA, MikroTik RouterOS) point to sustained attacker focus on widely deployed platforms and edge devices. The convergence of AI-assisted attacks (ransomware in under 10 hours, AI coding agent exploitation via GitSpawn) with traditional social engineering (ClickFix variants targeting browsers, Microsoft 365 vishing) shows attackers adapting both automation and human manipulation tactics. Cloud misconfigurations remain endemic across all major providers, with distinct risk profiles requiring provider-specific hardening. Post-compromise tooling like PEEP demonstrates attackers investing in persistence frameworks that blend into legitimate browser activity, complicating detection.