StyleSmuggler CVE-2026-75650 CVSS 10.0: four days active exploitation, emergency hotfix APSB26-146 now available, disable GraphQL immediately, rotate encryption key and all derived credentials      PaperCut education sector: Arctic Wolf confirms post-exploitation includes registry hive theft, Metasploit payloads, reconnaissance across K-12 and universities in US and Europe      Chrome CVE-2026-85046: V8 type confusion zero-day on CISA KEV, sixth Chrome zero-day in 2026, update to 152.0.7977.82 now      StyleSmuggler CVE-2026-75650 CVSS 10.0: four days active exploitation, emergency hotfix APSB26-146 now available, disable GraphQL immediately, rotate encryption key and all derived credentials      PaperCut education sector: Arctic Wolf confirms post-exploitation includes registry hive theft, Metasploit payloads, reconnaissance across K-12 and universities in US and Europe      Chrome CVE-2026-85046: V8 type confusion zero-day on CISA KEV, sixth Chrome zero-day in 2026, update to 152.0.7977.82 now     
CyberSipTM
Intelligence without the noise
Issue No. 125
September 8, 2026
3 items · past 4 days
<5 min read
Holiday weekend picture

Adobe released an emergency hotfix for CVE-2026-75650 on September 7 at CVSS 10.0, addressing the StyleSmuggler zero-day in Magento Open Source and Adobe Commerce that has been actively exploited since September 4 and backdoored fully-patched stores by poisoning PHP code through a failed payment email and executing it through Magento's template system without any authentication, with Adobe's advisory for APSB26-146 explicitly recommending that all affected merchants rotate their encryption key and every credential that key protected, including admin passwords, REST, SOAP, and GraphQL integration tokens, OAuth client secrets, payment gateway API keys, database credentials, SSH keys, deploy keys, and third-party extension API keys. Arctic Wolf confirmed over the weekend that PaperCut CVE-2026-81578 and CVE-2026-82078 have been used in sustained attacks targeting the education sector across K-12 schools and major universities in the United States and Europe, with post-exploitation activity including Windows registry hive collection tools, Metasploit Meterpreter Java payloads, and systematic commands to enumerate hosts, users, processes, and sensitive configuration data. Google released an emergency Chrome update for CVE-2026-85046, a V8 JavaScript engine type confusion zero-day confirmed exploited in the wild, with CISA adding it to the Known Exploited Vulnerabilities catalog the same day the fix shipped in Chrome 152.0.7977.82 across all platforms, the sixth Chrome zero-day fixed in 2026.

Holiday weekend intelligence
3 items
01 CriticalStyleSmugglerMagento / Adobe Commerce
StyleSmuggler CVE-2026-75650: fully-patched Magento and Adobe Commerce stores were backdoored for four days before an emergency hotfix shipped at CVSS 10.0
Attacks started September 4. Sansec published early on September 5 because stores were being compromised while they wrote the advisory. Adobe released APSB26-146 on September 7. Apply the VULN-39341 hotfix immediately. Then rotate the Magento encryption key and every credential derived from it. Being on the latest patch level did not prevent compromise: the first confirmed victim was running 2.4.6-p15, the highest patch level Adobe shipped for that branch.
CVECVE-2026-75650
CVSS 10.0
AdvisoryAPSB26-146
September 7, 20:20 UTC
Priority 1
ExploitationStarted Sept 4
No CVE gap: 3 days
No patch: 3 days
AffectedAll versions
2.4.4 through 2.4.9
Incl. latest patches
HotfixVULN-39341
composer patch
via repo.magento.com
Interim mitigationDisable GraphQL
Dutch e-commerce security firm Sansec discovered a zero-day vulnerability in Magento Open Source and Adobe Commerce and named it StyleSmuggler. Active exploitation began on September 4, 2026. Sansec published its advisory on September 5 specifically because stores were being compromised in real time and the company did not want to wait for coordinated disclosure to complete while active attacks were underway. Adobe published emergency advisory APSB26-146 on September 7 at 20:20 UTC, assigned CVE-2026-75650, rated the flaw CVSS 10.0 with Priority 1, its highest severity classification, and released a hotfix. StyleSmuggler is a two-stage attack that does not require authentication. In the first stage, the attacker injects PHP code into a location that Magento's logging subsystem will later write to. Sansec's name reflects the injection path: the flaw specifically leverages the styles properties of Magento's template system to smuggle malicious code past the application's validation logic. A failed payment transaction email report is one of the documented injection vectors, as the payment failure handler writes attacker-controlled content to a log or queue path without adequate sanitization. In the second stage, the attacker triggers Magento to execute the poisoned code by causing the template system to process the logged content. The result is arbitrary code execution in the context of the web server without any authentication requirement and without any action from a store administrator or customer. A successful attack installs a persistent Linux backdoor on the server, typically disguised as a kernel process using a cron job, and may operate outside the Magento web root, making simple web-accessible file scans inadequate for detecting the implant. Sansec confirmed that all current Magento Open Source versions from 2.4.4 through 2.4.9 are affected and that it reproduced the full unauthenticated exploit chain on clean installations of 2.4.7, 2.4.8, and 2.4.9. The first confirmed victim was running Magento 2.4.6-p15 with Adobe's July and August 2026 security updates applied, which represents the highest patch level Adobe ships for the 2.4.6 release line. Being fully current on published patches did not prevent compromise. Adobe's emergency hotfix ships as a composer patch: download VULN-39341-composer-patches.zip from repo.magento.com and apply it. Confirm installation with the magento-patches status command. Adobe's advisory explicitly recommends that all merchants rotate their encryption key and every credential that key protects after applying the hotfix, covering admin passwords, REST, SOAP, and GraphQL integration tokens, OAuth client secrets, payment gateway API credentials, database credentials, SSH keys, deploy keys, and third-party extension API keys. Until the hotfix is applied, Sansec's recommended interim mitigation is to disable GraphQL, which removes one of the attack paths StyleSmuggler uses to reach the injection vector.
Magento and Adobe Commerce run hundreds of thousands of online stores worldwide, handling payment transactions, customer personal data, and order history. The specific credential rotation Adobe is recommending reflects what is at risk in the Magento encryption architecture: the store's encryption key is the root key from which admin credentials, payment processor API keys, and database credentials are derived or protected. An attacker with code execution through StyleSmuggler has access to the Magento file system, including the local.xml or env.php configuration files that contain the encryption key and database credentials in the cleartext or encrypted form. With the key and credentials, an attacker can decrypt stored customer payment tokens, access the database directly, and maintain access to connected payment and shipping services even after the backdoor is removed from the server. The four-day exploitation window before the hotfix was available, combined with the fact that being fully current on patches provided no protection, means any Magento or Adobe Commerce store that was internet-accessible during September 4-7 should be treated as having been in the exploitation window.
Adobe's APSB26-146 advisory notes that the hotfix was tested against the 2026-aug releases of Adobe Commerce 2.4.4 through 2.4.9, Magento Open Source 2.4.4 through 2.4.9, and Adobe Commerce B2B 1.3.3 through 1.5.3. Older versions in those branches are affected but the patch is unverified for them. The advisory does not address Adobe Commerce on Cloud separately, meaning hosted cloud merchants should verify with Adobe whether the hotfix has been applied to their cloud environment or whether manual action is required. Sansec published a companion scanner, eComscan, that can detect the StyleSmuggler implant and associated backdoor artifacts. The persistent Linux backdoor typically runs outside the Magento web root and may survive a reinstallation of Magento itself if the underlying operating system and cron jobs are not also cleaned. Post-exploitation investigation should extend to the operating system level, not only to the Magento application directory. Disrex Group published a public incident-response repository on September 5 with specific indicators of compromise and mechanism details from two confirmed victims.
  • Apply Adobe's emergency hotfix VULN-39341 immediately. Download VULN-39341-composer-patches.zip from repo.magento.com, apply it as a composer patch to the Magento installation, and confirm proper installation by running vendor/bin/magento-patches -n status and verifying the 39341 patch shows as applied. Adobe Commerce on Cloud merchants should verify with Adobe whether the patch has been applied automatically or requires manual action.
  • Disable GraphQL as an interim control if the hotfix cannot be applied immediately. This removes one of the documented attack paths for StyleSmuggler. Disabling GraphQL may affect storefront features that rely on it; assess impact before disabling in production and treat this as a temporary measure until the hotfix is confirmed applied.
  • After applying the hotfix, rotate the Magento encryption key and all credentials that key protects, as Adobe explicitly recommends in APSB26-146. This covers: the Magento admin password and all other admin account passwords, REST, SOAP, and GraphQL API integration tokens, OAuth client secrets, payment gateway API credentials for all connected payment processors, database credentials, SSH keys and deploy keys, and API keys for any third-party extensions. If the store was accessible during September 4-7, treat the pre-hotfix period as a confirmed exposure window and prioritize immediate rotation over next-cycle rotation.
  • Run Sansec's eComscan or review the Disrex Group incident-response repository indicators to scan for the StyleSmuggler implant and associated Linux backdoor artifacts. Check cron jobs on the web server operating system for unexpected entries, review processes running as the web server user for anything unexpected, and look for files in directories outside the Magento web root that match the backdoor signatures Sansec and Disrex Group documented.
Fully patched stores were compromised. The first victim had applied every available Adobe update. Four days of active exploitation before the emergency hotfix shipped at CVSS 10.0. Apply VULN-39341 now. Disable GraphQL if you cannot patch immediately. Then rotate the encryption key and every credential it covers, including payment gateway API keys. Then scan for the Linux backdoor. Patching alone is not remediation if the implant is already installed.
02 HighPaperCutEducation Sector
PaperCut exploitation in the education sector is confirmed: Arctic Wolf observed registry hive collection, Meterpreter payloads, and systematic reconnaissance across K-12 and universities
This updates Issues 119, 121, and 122. Arctic Wolf identified sustained post-exploitation activity targeting education sector PaperCut servers including Windows registry hive collection tools consistent with credential harvesting, Metasploit Meterpreter Java payloads for persistent remote access, and enumeration commands covering hosts, users, processes, and sensitive configuration data. The targeting of the education sector specifically continues the ransomware affiliate pattern from the 2023 PaperCut exploitation cycle.
SourceArctic Wolf
Adversary Research
Targets confirmedK-12 schools
Major universities
US and Europe
Post-exploitationRegistry hive
collection tools
Metasploit/Meterpreter
Java payloads
Host/user/process
enumeration
CVEsCVE-2026-81578
CVE-2026-82078
CISA KEV
Patch statusApply Emergency
Patch Release 2
Known bypasses exist
in second patch
Arctic Wolf's Adversary Research Team published findings over the weekend confirming that threat actors are actively exploiting CVE-2026-81578 and CVE-2026-82078, the PaperCut authentication bypass and remote code execution chain, in sustained attacks targeting education sector organizations. Arctic Wolf observed exploitation activity across K-12 school districts and major universities in the United States and Europe. The post-exploitation activity Arctic Wolf documented goes substantially further than the web shell deployment observed in the initial August 28 incidents. The confirmed post-exploitation tools and techniques include Windows registry hive collection tools, specifically programs designed to extract copies of the Security Account Manager, SYSTEM, and SECURITY registry hives, which contain Windows credential material including password hashes for local accounts. The presence of registry hive collection tools indicates the attackers are specifically targeting credential material from the PaperCut server and any Windows systems reachable from it for use in lateral movement or offline cracking. Arctic Wolf also observed delivery of Metasploit Meterpreter-related Java payloads through the PaperCut server, establishing a persistent remote access capability through the Meterpreter framework rather than a simple web shell. Java Meterpreter payloads delivered through a Java-based application server like PaperCut can be difficult to detect with endpoint security tools that do not specifically cover Java runtime execution. The post-exploitation reconnaissance commands Arctic Wolf documented covered host identification including hostname and IP enumeration, user account enumeration, running process listing, and collection of sensitive configuration data from the PaperCut server including database connection strings and credentials stored in PaperCut's configuration files.
The education sector targeting continues the pattern from the 2023 PaperCut exploitation wave, where Bl00dy ransomware specifically targeted K-12 schools and universities using CVE-2023-27350 following a joint CISA and FBI advisory. The 2023 wave converted PaperCut server access into ransomware deployment across multiple educational institutions. The registry hive collection tools Arctic Wolf observed are the tooling that precedes lateral movement and persistence across a Windows domain: an attacker who extracts the SAM hive from a PaperCut server that is joined to the school or university Active Directory domain has credential material that can be used to authenticate to other domain systems. The Meterpreter Java payloads provide persistent remote access that survives server reboots and can be used to execute additional payloads after any investigative response that focuses only on removing visible artifacts. Organizations that applied Emergency Patch Release 2 but did not conduct a post-exploitation review are operating on the assumption that the patch eliminated active attacker access, which the Arctic Wolf findings indicate may not be true: attackers who established Meterpreter persistence before or after the patch may still have active access through that channel.
The registry hive collection technique Arctic Wolf observed is operationally significant for domain-joined PaperCut servers. PaperCut Application Servers are frequently domain-joined Windows machines, which gives local administrator access on the PaperCut server the ability to extract credential material from the SYSTEM and SECURITY hives that includes the machine account password hash. That hash can be used in pass-the-hash attacks against other domain members or in Kerberos-based lateral movement. Any domain-joined Windows PaperCut server that was running a vulnerable version and was accessible to exploitation should be treated as having potentially had its domain credential material extracted. The follow-on action is not only to patch PaperCut but to assess whether the domain account associated with the PaperCut server's machine account needs to be reset and whether privileged accounts that have authenticated to or from the PaperCut server during the exploitation window need to have their credentials rotated.
  • Conduct a post-exploitation review of all PaperCut Application Servers that were running vulnerable versions before Emergency Patch Release 2 was applied. Specifically look for registry hive extraction activity in Windows security event logs, specifically Event IDs 4656, 4663, and 4660 for SAM, SYSTEM, and SECURITY hive accesses, and for Java processes spawned by the PaperCut application process that are not consistent with normal PaperCut operations. The presence of Meterpreter-related Java payloads may require scanning for specific Java class names or network connections characteristic of Meterpreter's staging protocol.
  • For domain-joined PaperCut servers that were accessible before Emergency Patch Release 2, assess whether machine account credentials and any privileged domain accounts that authenticated to the PaperCut server during the exploitation window need to be rotated. The registry hive collection tooling Arctic Wolf observed is specifically designed to extract this material. Coordinate with Active Directory administrators to identify accounts that have logged on to or from the PaperCut server and evaluate whether they require password resets.
  • Verify that Emergency Patch Release 2 is applied and that web access to the PaperCut Application Server is restricted to trusted IP ranges. Monitor PaperCut's advisory for a third patch release addressing the WatchTowr-identified bypasses of the second patch, and apply it immediately when available. The combination of a network access control and the latest available patch is the current best defensive posture given the confirmed bypass status of the first patch and the pending bypass status of the second.
Registry hive collection. Meterpreter Java payloads. Systematic reconnaissance across K-12 and universities. This is not web shell dropping; this is the credential harvest and persistence phase that precedes ransomware. The 2023 PaperCut exploitation wave ended in school districts locked out of their systems. Review the Windows event logs for SAM hive access events. Scan for Java Meterpreter activity. If the PaperCut server is domain-joined and was exposed, assess whether domain credentials need rotating before this goes further.
03 HighChrome CVE-2026-85046CISA KEV
Chrome's sixth zero-day of 2026 is on CISA KEV and was confirmed exploited in the wild: update to 152.0.7977.82 on all platforms now
CVE-2026-85046 is a V8 type confusion flaw that lets a remote attacker execute arbitrary code inside the Chrome sandbox via a crafted web page. The exploit exists in the wild. CISA added it the same day Google shipped the emergency fix. This is the sixth Chrome zero-day in 2026, which is running at twice the pace of prior years. Enterprise admins should not wait for staged rollout; accelerate deployment via managed policy today.
CVECVE-2026-85046
CVSS 8.8
TypeV8 type confusion
CWE-843
Sandbox RCE
Fixed inChrome 152.0.7977.82
(Windows, macOS)
.82 for Linux
KEV addedSeptember 5, 2026
Same day as patch
2026 zero-days6th Chrome
zero-day in 2026
Google released an emergency Chrome security update on September 5, 2026, patching CVE-2026-85046, a type confusion vulnerability in the V8 JavaScript and WebAssembly engine. Security researcher Salvatore Gulizia reported the vulnerability to Google on August 4, 2026, and Google confirmed an exploit exists in the wild. Type confusion vulnerabilities in V8 occur when the engine treats a JavaScript object as the wrong type, which can be leveraged to read or write memory outside intended boundaries, enabling attackers to achieve arbitrary code execution inside Chrome's renderer sandbox through a crafted HTML page that any user who visits it would trigger automatically, with no additional interaction required. The emergency fix ships in Chrome 152.0.7977.82 and 152.0.7977.83 for Windows and macOS and 152.0.7977.82 for Linux. CISA added CVE-2026-85046 to its Known Exploited Vulnerabilities catalog on September 5, the same day the patch was released, confirming federal civilian agencies must update Chrome immediately. CVE-2026-85046 is the sixth Chrome zero-day patched in 2026, a rate of approximately one every six weeks that represents roughly twice the pace of prior years. The prior five Chrome zero-days in 2026 spanned V8 type confusion, heap use-after-free, and sandbox escape classes. Google's emergency release was made available to Chrome's staged rollout infrastructure on September 5, which means some users will receive it automatically within hours while others in later rollout stages may not receive it for several days without manual action.
Browser zero-days confirmed exploited in the wild are drive-by attack vectors: a user who visits a malicious or compromised page in an unpatched browser triggers code execution without any further action. Chrome's installed base of hundreds of millions of endpoints means that any confirmed in-the-wild Chrome zero-day represents a potential attack surface against a very large population during the window between patch release and universal update coverage. Type confusion vulnerabilities in V8 are particularly consistent exploitation targets because V8 handles the JavaScript execution for every web page Chrome renders, making the attack surface both universal and browser-internal. Six Chrome zero-days in a single year at the pace observed in 2026 is a data point consistent with increased attacker focus on browser exploitation as a primary initial access vector, which aligns with the broader pattern of exploitation shifting toward the browser layer as perimeter defenses improve.
Chrome's staged rollout means that enterprise endpoints managed through group policy or mobile device management may not receive the update automatically within the timeframe that the CISA KEV urgency implies. Enterprise administrators should verify that the minimum Chrome version policy is set to 152.0.7977.82 or higher and push an accelerated update through their endpoint management infrastructure rather than relying on Chrome's background update mechanism. The update can be verified at chrome://settings/help; the version string should show 152.0.7977.82 or higher after a browser restart. The staged rollout cadence means that checking one machine does not confirm that all machines in a fleet are updated: centralized fleet version reporting through the management console is the appropriate verification method for enterprise environments. Endpoints that have not yet received the update through staged rollout but are covered by an enterprise update policy should have the policy-forced update confirmed before the end of today given the CISA KEV status.
  • Update Chrome to version 152.0.7977.82 or higher on all endpoints immediately. For individual machines, open Chrome and navigate to chrome://settings/help to trigger an immediate update check and confirm the installed version after restarting. For enterprise environments, push the update through your endpoint management infrastructure and set the minimum required Chrome version to 152.0.7977.82 via group policy or MDM configuration rather than waiting for staged rollout to reach all endpoints.
  • Verify that Chrome is set to apply updates automatically on all managed endpoints. CVE-2026-85046 is the sixth Chrome zero-day in 2026, and prior Chrome zero-days this year have also been confirmed exploited in the wild. An enterprise Chrome update policy that ensures prompt deployment of emergency patches is a standing control that covers future Chrome zero-days as well as this one.
Sixth Chrome zero-day this year. V8 type confusion, sandbox RCE, exploit confirmed in the wild, CISA KEV the same day as the patch. Update to 152.0.7977.82. Do not wait for staged rollout in enterprise environments: push it via managed policy today. At one confirmed zero-day every six weeks, Chrome is not an intermittent concern in 2026. It is an ongoing patching priority.
Cross-source standouts
01
StyleSmuggler and the patching-without-protection problem: when being fully current on vendor patches is not enough
The first confirmed StyleSmuggler victim was running Magento 2.4.6-p15 with Adobe's July and August 2026 security updates applied. That is the highest patch level Adobe ships for the 2.4.6 release line. The store was, by every standard metric, up to date. It was still compromised on September 4, the first day of active exploitation. This reflects a structural reality in e-commerce platform security that this brief has tracked in several contexts this year. Patch management frameworks are designed to address known vulnerabilities through the vendor's published patch cycle. They do not address zero-day vulnerabilities by definition, because zero-days are vulnerabilities for which no patch exists at the time of exploitation. For e-commerce platforms specifically, the risk profile is particularly concentrated: these platforms handle payment transactions, store customer PII, and are connected to payment processors and fulfillment systems through API credentials that provide direct access to financial infrastructure. A zero-day in Magento or Adobe Commerce is not only a web server compromise; it is a path to the payment gateway credentials, the customer database, and the order management system in a single exploitation chain. Adobe's explicit recommendation to rotate every credential the encryption key protects reflects how far that access extends. The interim mitigation Sansec recommended, disabling GraphQL, is notable because it removes an attack path rather than fixing the underlying flaw. For e-commerce operators who cannot apply the hotfix immediately, understanding which specific attack surface to remove as a temporary control is more operationally useful than generic hardening advice. If a specific attack path depends on a specific feature being enabled, that feature can be temporarily disabled to shrink the exposed surface while the patch is prepared. This is the same logic as disabling open registration on Gitea from Issue 117: remove the specific precondition the exploit depends on before the full fix is available.
02
The holiday weekend exploitation window: why the September 4-8 period is specifically attractive for attackers targeting enterprise and institutional infrastructure
This brief covers issues Monday through Friday and took a four-day break for the Labor Day holiday weekend. Three of today's stories have a specific relationship to that break. StyleSmuggler exploitation began September 4 and ran for four days before the emergency hotfix shipped September 7. The PaperCut education sector attacks have been running continuously through the weekend and are now documented with post-exploitation tooling consistent with credential harvesting and ransomware preparation. Chrome's zero-day was confirmed exploited in the wild and CISA KEV'd on September 5, the Friday before a long weekend. This timing is not coincidental in the broadest pattern. Security researchers at Secureworks, Mandiant, and Sophos have documented that ransomware groups and nation-state actors preferentially time major exploitation campaigns and ransomware deployment for long holiday weekends, when staffing levels in enterprise and institutional security operations centers are reduced and response times are extended. The PaperCut education sector targeting is particularly consistent with this: schools and universities have minimal IT security staff over Labor Day weekend, the exploitation campaign had already been active for ten days, and the post-exploitation tooling Arctic Wolf documented suggests the credential harvesting and lateral movement phase was underway through the weekend. The operational implication for security teams returning from the holiday is not only to review the three items in today's issue but to assess whether any of the systems tracked in the Still Watching section below showed unusual activity during the September 4-8 window when monitoring coverage may have been reduced.
Still watching
Days 5–7+
SonicWall SMA1000 CVE-2026-83548 / CVE-2026-83549 (Issue 123 · zero-day chain, confirmed exploitation, no IoCs published) — apply hotfixes 12.4.3-03526 or 12.5.0-02952. Re-image if compromise is suspected given absence of indicators and history of persistent malware from prior SMA1000 campaigns. Review VPN and admin logs from before the hotfix date.
Day 7+
JFrog Artifactory CVE-2026-82329 (Issue 122 · CVSS 9.8, active exploitation, admin tokens being created) — update self-hosted instances to 7.161.20 or a patched branch version. Review Artifactory audit logs for unexpected admin token creations after August 28. JFrog SaaS was patched before disclosure.
Day 7+
ShieldBreak CVE-2026-69414 (Issue 113 · Defender patch bypass, patch in progress per August 21) — low privilege to SYSTEM on fully patched Windows 10, 11, and Server 2025. No patch yet. Monitor MSRC and apply the day it ships. Verify endpoint detection is current for CVE-2026-69414 specifically.
Day 7+
GitLab CVE-2026-19478 (Issue 111/114 · confirmed exploited, NSA/CISA joint advisory, 90-day embargo through mid-November 2026) — patch available for 19.x and 18.11 branches. Versions 18.2 through 18.10 have no patch available through mid-November. Restrict /api/graphql endpoint and hunt @gl_introduced in web server logs.
Day 7+