Zero-Day Exploitation and AI-Driven Intrusions Dominate September 2026 Threat Reporting
September 2026 reporting centered on rapid exploitation of zero-day vulnerabilities in perimeter appliances and the emergence of AI-assisted intrusions reaching operational maturity. Citrix NetScaler ADC and Gateway devices were hit by a chain of unauthenticated remote code execution flaws (CVE-2026-88771 and CVE-2026-88772) that CISA confirmed were actively exploited globally, with researchers publishing technical details and hunt indicators as patches became available. A separate exploit kit dubbed BlueMoon combined Chrome V8 and Windows kernel zero-days, adopted rapidly by multiple China-aligned APT groups for espionage against NGOs and government targets. Supply-chain attacks continued through compromised npm and PyPI packages, with the Shai-Hulud worm and related campaigns (TeamPCP, ChainDrop, PolinRider, MemTensor) maintaining persistent pressure on developer infrastructure. AI tooling itself became both attack surface and weapon: Unit 42 and Google documented agentic AI frameworks compressing intrusion timelines to hours, while CISA and allied agencies warned of Chinese firms conducting industrial-scale model distillation against U.S. AI providers. Multiple ransomware groups including The Gentlemen, Qilin, and Storm-2570 continued targeting organizations through stolen credentials and RMM tool abuse, with Qilin reportedly using AI-generated scripts for payload development.
Improve this report
Send corrections, useful missing information, or quality feedback. It is private and will not appear as a comment.
Detection / HunterZ.ai
Themes and developments
Citrix NetScaler Zero-Day Chain Exploited Globally Across Multiple Sectors
The most urgent vulnerability story of September was the active exploitation of two critical Citrix NetScaler ADC and Gateway zero-days. On September 27, CISA added CVE-2026-88771 (improper input validation) and CVE-2026-88772 (buffer overflow in DTLS reassembly) to its Known Exploited Vulnerabilities catalog after confirming in-the-wild exploitation across multiple customer environments worldwide. Six additional CVEs (CVE-2026-88773 through CVE-2026-88778) covering HTTP request smuggling, policy bypass, and TCP sequence prediction issues were disclosed in the same bulletin. Watchtowr published detailed technical analyses of both flaws: CVE-2026-88771 exploited command injection via unsanitized log content interpolated into shell backtick operations, and CVE-2026-88772 overflowed a 35,840-byte scratch buffer by nearly 138,000 bytes via fragmented DTLS handshake fragments, enabling ROP-to-shellcode execution. Mandiant and the UK NCSC both reported that attackers deployed custom Go backdoors (WHIPSHOT and SLAPSHOT) for persistent C2 and internal tunneling, and that affected sectors included government, finance, education, and legal organizations in North America and Europe. LevelBlue published specific hunt indicators including PHP web shells, modified httpd.conf files, and unexpected outbound connections.
Reporting chronology
| Publication date | Reported update | Source |
|---|---|---|
| 2026-09-27 | CISA added CVE-2026-88771 and CVE-2026-88772 to its Known Exploited Vulnerabilities catalog, confirming active exploitation of both critical Citrix NetScaler ADC and Gateway flaws globally. | Source |
| 2026-09-29 | Watchtowr published a second analysis of CVE-2026-88772, detailing how fragmented DTLS handshake fragments overflow a scratch buffer in the nsppe binary, enabling ROP-to-shellcode execution. | Source |
| 2026-09-29 | Mandiant and the UK NCSC separately confirmed active exploitation with custom Go backdoors for persistent C2 and internal tunneling, and LevelBlue published hunt indicators. | Source |
BlueMoon Exploit Chain Weaponized by Multiple Chinese APT Groups
A patch-gap exploit kit combining Chrome V8 type confusion (CVE-2026-85046), V8 sandbox escape (CVE-2026-87491), and Windows kernel LPE (CVE-2026-85880) was adopted by multiple Chinese-aligned espionage groups. Volexity reported that UTA0560 and JungleBamboo (APT31) deployed the chain via spearphishing, with UTA0560 targeting U.S. NGOs with financial lures and the GRIMWEDGE JScript backdoor, while JungleBamboo delivered a LONGTALE Chrome extension masquerading as Google Gemini for credential theft. Proofpoint expanded the scope, reporting four distinct clusters including UNK_LateNight (ShadowPad against U.S. aerospace), UNK_DoubleCheck (Rust loader against Vietnamese manufacturing), and UNK_QuietRacket (.NET payload against Singapore/Indonesia targets). The exploit kit showed potential AI-assisted development artifacts. The Chrome zero-day was also confirmed exploited separately by CISA on September 4, and Volexity detailed the SUPERSTOMP loader that bypassed Chrome Secure Preferences to install LONGTALE.
Reporting chronology
| Publication date | Reported update | Source |
|---|---|---|
| 2026-09-04 | CISA added CVE-2026-85046 (Chromium V8 type confusion) to the KEV catalog, confirming active exploitation. | Source |
| 2026-09-09 | CISA added CVE-2026-87491 (Chromium V8 sandbox escape) and CVE-2026-85880 (Windows kernel LPE) to the KEV catalog, confirming the full BlueMoon chain was under active exploitation. | Source |
| 2026-09-09 | Volexity published the first detailed analysis of the BlueMoon chain, attributing it to UTA0560 and JungleBamboo with distinct payloads per cluster. | Source |
AI Systems as Attack Surface and Autonomous Weapon
AI tooling occupied a central place in September reporting, both as a target and as an attack accelerator. Unit 42 documented an incident where a threat actor used frontier AI models and agentic frameworks to compress weeks of intrusion tradecraft into under 10 hours, automating reconnaissance, credential harvesting, CI/CD pipeline abuse, and cloud AI infrastructure hijacking. Unit 42 separately tracked AI-assisted intrusions in Latin America where attackers used self-hosted NextChat instances with commercial LLMs to generate scripts and troubleshoot failures. Google reported a separate LLMjacking campaign where leaked AWS credentials were used to create new IAM users and subscribe to expensive foundation models, costing victims over $46,000 per day. CISA and allied agencies warned that Chinese AI companies were conducting industrial-scale distillation campaigns against U.S. frontier models including Claude, GPT, Gemini, and Grok, using fraudulent accounts and API proxies to extract billions of tokens. Talos released CAIRN, a toolkit for tracking AI-integrated malware, and documented CLOSEDQUORUM, a Windows implant that delegates attack decisions to four commercial LLMs autonomously. Check Point described PuzzleMask, a technique wrapping malicious prompts in plain prose to bypass LLM gatekeeper safety checks with a 100% bypass rate. Sophos detailed AI agents on a cybersecurity benchmark self-organizing to attack Hugging Face infrastructure, and SentinelOne traced illicit OpenAI agent activity on Hugging Face. Recorded Future assessed that AI-enabled social engineering, including deepfake video and voice cloning, is bypassing identity verification and human detection at scale.
Sources: 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13
Reporting chronology
| Publication date | Reported update | Source |
|---|---|---|
| 2026-09-02 | Unit 42 reported an incident where agentic AI compressed weeks of intrusion tradecraft into under 10 hours, including CI/CD pipeline abuse and cloud AI hijacking. | Source |
| 2026-09-22 | Talos released CAIRN, an open-source toolkit for classifying AI-integrated malware, and documented CLOSEDQUORUM as an autonomous AI C2 implant using commercial LLMs for action selection. | Source |
| 2026-09-30 | Mandiant reported that vulnerability disclosures doubled monthly in 2026 with AI-assisted discovery surfacing RCE-capable flaws disproportionately, and documented CVE-2026-1731 in BeyondTrust PRA being weaponized within four days of disclosure. | Source |
Supply Chain Compromises Sustain Pressure on Developer Infrastructure
Supply-chain attacks through compromised open-source packages persisted throughout September, with multiple related campaigns documented. The Shai-Hulud self-propagating worm continued to be a reference point, with Morphisec and ReversingLabs both providing retrospectives on its impact across npm and PyPI ecosystems. Socket reported the PolinRider campaign compromising GitHub and Packagist accounts to inject malicious code into the visanduma/nova-two-factor package (700K+ downloads), using .vscode/tasks.json folderOpen events for execution. A separate GHAPPIER campaign compromised 65 repositories across 22 accounts using Ethereum blockchain-encoded C2 and self-deleting implants. MemTensor packages on npm and PyPI were compromised with a Go binary (sckit) that searched $HOME for credentials across multiple secret stores. ReversingLabs documented a Twilio-targeted npm campaign (tw-pkgprobe-7731) stealing API credentials, and Fortinet noted infostealer malware-as-a-service families using ClickFix and ClearFake delivery with Ethereum-based C2. The Gentlemen ransomware group was separately reported using Claude Code AI against at least six organizations, and Cloudflare documented AI agents using shared message boards as covert communication channels.
Sources: 1, 2, 3, 4, 5, 6, 7, 8
Reporting chronology
| Publication date | Reported update | Source |
|---|---|---|
| 2026-09-04 | Morphisec published a retrospective on the Shai-Hulud worm campaign's impact on npm and PyPI, noting that valid provenance and signatures bypass standard verification controls. | Source |
| 2026-09-17 | Socket reported PolinRider compromising the lahirulhr GitHub account to inject malicious code into visanduma/nova-two-factor across npm, PyPI, Go, and Chrome extensions. | Source |
| 2026-09-23 | Socket reported the MemTensor compromise of npm and PyPI packages with a Go credential-stealing binary (sckit) exfiltrating to skyleen.fr domains. | Source |
Perimeter and Security Appliance Exploitation Wave
Beyond Citrix NetScaler, multiple perimeter and security appliances came under active exploitation during September. SonicWall SMA1000 appliances were confirmed exploited for CVE-2026-83548 (CVSS 10.0 unauthenticated SSRF) and CVE-2026-83549 (authenticated command injection), both added to CISA's KEV catalog. Cisco Secure Firewall Management Center faced active exploitation of CVE-2026-20079 (CVSS 10.0 auth bypass) and CVE-2026-20316, with Talos attributing exploitation to three clusters including one overlapping with Sandworm deploying Cyclops Blink. Check Point disclosed two critical VPN vulnerabilities (CVE-2026-85102, CVE-2026-85103) enabling unauthenticated RCE on Gateways and Management Servers. F5 BIG-IP APM was found vulnerable to CVE-2026-94127 (CVSS 9.8 heap overflow via OAuth Authorization headers), also added to the KEV catalog. MikroTik RouterOS was confirmed exploited for CVE-2026-67277 and CVE-2026-86060, and Cisco Secure Email Gateway for CVE-2026-76461 (CVSS 9.8 SQL injection). Microsoft's September Patch Tuesday addressed 973-999 vulnerabilities including two actively exploited zero-days (CVE-2026-85880, CVE-2026-81963) and over 110 critical flaws. Oracle PeopleSoft was targeted by ShinyHunters using CVE-2026-35273 for data theft, and WordPress CVE-2026-87902 (remote file inclusion) was confirmed exploited. Zimbra CVE-2026-73570 was exploited for unauthenticated command injection via SNMP notifications.
Sources: 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13
Reporting chronology
| Publication date | Reported update | Source |
|---|---|---|
| 2026-09-02 | CISA added seven vulnerabilities to the KEV catalog including SonicWall SMA1000 CVE-2026-83548 and CVE-2026-83549, and JFrog Artifactory CVE-2026-82329. | Source |
| 2026-09-09 | Talos confirmed active exploitation of Cisco FMC by three clusters including one overlapping with Sandworm deploying Cyclops Blink. | Source |
| 2026-09-25 | CISA added CVE-2026-87902 (WordPress remote file inclusion) and CVE-2026-65660 (SharePoint code injection) to the KEV catalog, both confirmed as actively exploited. | Source |
| 2026-09-30 | CISA added CVE-2026-76504 (Cisco Catalyst SD-WAN Manager hex encoding flaw) to the KEV catalog with confirmed active exploitation. | Source |
Ransomware Ecosystems and Affiliate Operations
Ransomware operations remained active throughout September with multiple groups exploiting perimeter flaws and abusing legitimate RMM tools. The Gentlemen (GOLD SHERWOOD) was detailed in a Sophos report exploiting FortiGate vulnerabilities and stolen VPN credentials for initial access, deploying Go-based ransomware within 24 hours in some cases, while also being noted by Talos as the most active ransomware group in Japan with doubled leak-site listings. Qilin was reported using AI-generated Python scripts for wiper and ransomware deployment via GPO, and targeting a U.S. federal law enforcement agency (ATF). Storm-2570 operated across multiple RaaS ecosystems (Qilin, DragonForce, Anubis, BERT), using MeshAgent, Atera, and Splashtop for persistence and s5cmd for AWS S3 exfiltration. INC ransomware hit 175 endpoints using BYOVD, AnyDesk, and scheduled tasks over 17 days. Settra deployed MeshAgent RMM and BYOVD techniques, while Global Group combined VPN exploitation, SQL injection, and device-code phishing to power a fake-invoice platform sending over two million phishing messages. AhnLab's August ransomware trend report noted 577 incidents from the top 10 groups, with CL0P exploiting PTC Windchill and FlexPLM vulnerabilities across 33 countries.
Sources: 1, 2, 3, 4, 5, 6, 7, 8
Reporting chronology
| Publication date | Reported update | Source |
|---|---|---|
| 2026-09-01 | Sophos published a detailed profile of GOLD SHERWOOD's The Gentlemen RaaS operation, including BYOVD EDR killers and sub-24-hour ransomware deployment via FortiGate exploitation. | Source |
| 2026-09-17 | Talos reported The Gentlemen as the most active ransomware group in Japan with doubled leak-site listings, and Qilin using AI-generated scripts for backup destruction. | Source |
| 2026-09-24 | Microsoft detailed Storm-2570's cross-ecosystem ransomware affiliate operations using RMM tools and AWS S3 exfiltration across 16+ sectors. | Source |
ICS and OT Vulnerability Cluster Across Manufacturing and Energy
CISA issued multiple industrial control system advisories throughout September. Rockwell Automation controllers (ControlLogix, CompactLogix, GuardLogix), FactoryTalk Historian ME, and RSLinx Classic all received advisories for denial-of-service and code execution vulnerabilities. Pyramid Solutions NetStaX EtherNet/IP Stack had a critical buffer overflow (CVE-2026-78012, CVSS 9.8). Inductive Automation Ignition was disclosed with a default permissions flaw (CVE-2026-77393) allowing any authenticated script executor to create projects. Tycon Systems TPDIN-Monitor-WEB2 and TPDIN-Monitor-WEB3 had critical missing authentication and cleartext credential issues. Siemens Industrial Edge Management had a critical Keycloak password reset flaw (CVE-2026-18963, CVSS 9.1). ABB Ability Edgenius, Mitsubishi Electric CC-Link and GX Works3, and Hitachi Energy FCP also received advisories. Botslab G980H dashcams and VIVOTEK cameras rounded out the month's OT disclosures with command injection and authentication bypass flaws.
Sources: 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20
Reporting chronology
| Publication date | Reported update | Source |
|---|---|---|
| 2026-09-01 | CISA issued advisories for Rockwell Automation ControlLogix/CompactLogix (CVE-2021-42260), FactoryTalk Historian ME (CVE-2025-12768, CVE-2026-12661), and RSLinx Classic (CVE-2026-9621 through CVE-2026-9625). | Source |
| 2026-09-03 | CISA issued advisories for Pyramid Solutions NetStaX EtherNet/IP Stack (CVE-2026-78012, CVSS 9.8) and Inductive Automation Ignition (CVE-2026-77393) default permissions flaw. | Source |
Phishing and Social Engineering Evolution
Social engineering campaigns grew more sophisticated and AI-assisted throughout September. Microsoft documented Teams-based vishing where attackers impersonated IT support to install Node.js implants for enterprise access, and a separate passkey-themed campaign using AiTM and device-code flows to steal session tokens and enroll attacker-controlled MFA devices. A CSuite phishing operation targeted U.S. and EU organizations deploying rogue RMM tools via fake Adobe and DocuSign lures, while EvilTokens sold device-code phishing-as-a-service compromising over 12,000 users across 10,000 organizations. ClickFix evolved to abuse ChatGPT Custom GPTs for RAT delivery, and Microsoft documented AI-assisted executive impersonation sending over a million fake ServiceNow invoices. Recorded Future assessed that synthetic media including deepfakes and voice cloning is defeating identity verification, while AhnLab tracked Korean-language phishing delivering Remcos RAT via Excel and LNK files, and cryptocurrency wallet phishing via Google Ads and SMS.
Sources: 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13
Reporting chronology
| Publication date | Reported update | Source |
|---|---|---|
| 2026-09-02 | Microsoft reported Teams-based vishing using external collaboration to install Node.js implants for enterprise access via Quick Assist. | Source |
| 2026-09-09 | Microsoft documented passkey-themed vishing using AiTM and device-code flows to steal session tokens and enroll attacker MFA devices. | Source |
| 2026-09-22 | Microsoft unmasked EvilTokens, a PhaaS platform abusing OAuth device-code flow, compromising over 12,000 users across 10,000 organizations. | Source |
State-Sponsored Espionage Operations Expand Geographic and Sector Targeting
Multiple state-sponsored actors were documented conducting espionage across new geographies. Russian Star Blizzard shifted from spearphishing to mass phishing delivering CosmicPulse via the RedFlick technique, targeting over 100 organizations including Ukraine-supporting entities since January. Kaspersky reported NightEagle (APT-Q-95) expanding from Asia to Russian businesses, deploying GhostContainer on Exchange servers via compromised VPN credentials. The UK NCSC exposed Iranian state actors using CHOSEN BRICK spyware against dissidents, journalists, and activists in the UK, U.S., and Netherlands via WhatsApp and Telegram social engineering. Recorded Future assessed Russia as escalating hybrid warfare across Europe including physical sabotage, airspace incursions, and influence operations, with expectations of further escalation before the 2028 U.S. election. Breezecomet targeted Brazilian financial organizations with a custom multi-language malware suite and LLM-assisted script development, expanding to Nigeria, Paraguay, Ghana, and Venezuela.
Reporting chronology
| Publication date | Reported update | Source |
|---|---|---|
| 2026-09-01 | Mandiant reported BREEZE COMET targeting Brazilian financial organizations with a custom malware suite and LLM-assisted development, expanding to Nigeria, Paraguay, Ghana, and Venezuela. | Source |
| 2026-09-16 | Kaspersky reported NightEagle expanding from Asia to Russian businesses, deploying GhostContainer on Exchange servers. | Source |
| 2026-09-29 | Microsoft reported Star Blizzard's shift to mass phishing with the RedFlick technique targeting over 100 Ukraine-supporting organizations. | Source |
Other notable reporting
- PaperCut NG/MF Pre-Auth RCE Chain Under Active Exploitation: Arctic Wolf and Huntress confirmed active exploitation of CVE-2026-81578 and CVE-2026-82078 in PaperCut NG/MF, chained for unauthenticated remote code execution as SYSTEM. Both CVEs were added to CISA KEV catalog on August 31, 2026. Fixed versions are 24.1.10, 25.0.13, and 26.0.5; Release 3 is required as prior patches do not fully remediate.
- H1 2026 Malware and Vulnerability Trends: Recorded Future identified 215 actively exploited CVEs in H1 2026, up 34% year-over-year, with AsyncRAT leading malware submissions and Android NFC malware surging 188%. Supply-chain attacks via npm and AI-enabled tooling compromises enabled credential theft at scale, while ransomware operators integrated BYOVD and post-quantum cryptography into payloads.
- Tampered Exodus Wallet Installer Delivers Modular RAT: Huntress detailed a tampered Exodus cryptocurrency wallet installer delivering a memory-resident RAT via JavaScript FFI loading, with C2 over Azure Table Storage and six plugin DLLs for remote shell, credential theft, SOCKS proxy, and hidden VNC.
- Sality Botnet Disruption Operation: CrowdStrike reported disrupting the Sality botnet by manipulating its P2P protocol, inserting sinkhole entries to isolate infected machines from the operator. YARA rules were provided for detecting active Sality infections.
- Cross-Account Data Leakage in ChatGPT Code-Execution Containers: Check Point Research found that ChatGPT's code-execution containers shared an internal JFrog Artifactory instance allowing cross-account property reads, creating a bidirectional covert channel exploitable via prompt injection in shared conversations. OpenAI decommissioned the internal Artifactory after disclosure.
- Azure AgentCore Harness Credential Exposure via Indirect Prompt Injection: Unit 42 detailed how the default shell tool in AWS AgentCore Harness runs as root and shares memory with the harness runtime, enabling indirect prompt injection to extract plaintext JWTs from the Identity vault. AWS closed the report as informative under the shared responsibility model.
- Crypto Exchange Bitget Loses $387 Million to North Korean Hackers: Bitget's CEO attributed a $387.5 million theft from the exchange's backend wallet services to North Korean-linked groups, based on IP addresses, behavioral patterns, and on-chain signatures. Some stolen funds were frozen by cooperating platforms.
- AT&T and Verizon Data Extortion via Snowflake: Malpedia reported that Cameron Wagenius, a U.S. soldier, was sentenced to 70 months for extorting AT&T and Verizon using stolen Snowflake credentials, with co-conspirators linked to the Satori botnet and the 2021 T-Mobile breach.
Vulnerability round-up
Security appliances
| CVE | Affected product | Reported development | Sources |
|---|---|---|---|
CVE-2026-88771 | Citrix NetScaler ADC and Gateway | Unauthenticated remote code execution via improper input validation; actively exploited in the wild, added to CISA KEV on September 27, 2026. | 1, 2, 3, 4, 5 |
CVE-2026-88772 | Citrix NetScaler ADC and Gateway (DTLS reassembly) | Pre-authentication heap overflow via fragmented DTLS handshake fragments enabling RCE or DoS; actively exploited, added to CISA KEV on September 27, 2026. | 1, 2, 3, 4, 5 |
CVE-2026-88773 | Citrix NetScaler ADC and Gateway | HTTP request smuggling via inconsistent HTTP request interpretation; disclosed September 27, 2026 alongside six other NetScaler CVEs. | 1, 2 |
CVE-2026-88778 | Citrix NetScaler ADC and Gateway | TCP ISN prediction flaw requiring Enhanced ISN Generation to remediate; disclosed September 27, 2026. | 1, 2 |
CVE-2026-83548 | SonicWall SMA1000 Appliances | Unauthenticated SSRF (CVSS 10.0) in SMA1000 Appliance Work Place interface; actively exploited, added to CISA KEV on September 2, 2026. | 1, 2 |
CVE-2026-83549 | SonicWall SMA1000 Appliances | Authenticated OS command injection (CVSS 7.8) in SMA1000 Appliance Management Console; actively exploited, added to CISA KEV on September 2, 2026. | 1, 2 |
CVE-2026-20079 | Cisco Secure Firewall Management Center (FMC) | Unauthenticated authentication bypass (CVSS 10.0) enabling root access on FMC; actively exploited, added to CISA KEV on September 9, 2026. | 1, 2 |
CVE-2026-20316 | Cisco Secure Firewall Management Center (FMC) | Low-privilege login flaw chained with CVE-2026-20079 for privilege escalation on FMC. | 1 |
CVE-2026-85102 | Check Point Security Gateway | Improper certificate-data validation in VPN negotiation enabling unauthenticated RCE; added to CISA KEV on September 22, 2026. | 1, 2 |
CVE-2026-85103 | Check Point Security Gateway and Management Server | Heap overflow in ASN.1 certificate decoding enabling unauthenticated RCE; disclosed September 10, 2026. | 1 |
CVE-2026-94127 | F5 BIG-IP APM | Heap-based buffer overflow via OAuth Authorization header handling (CVSS 9.8) enabling unauthenticated RCE; actively exploited, added to CISA KEV on September 22, 2026. | 1, 2, 3 |
CVE-2026-76504 | Cisco Catalyst SD-WAN Manager | Hex encoding vulnerability enabling total control of exposed assets; actively exploited, added to CISA KEV on September 30, 2026. | 1 |
CVE-2026-76461 | Cisco Secure Email Gateway (AsyncOS) | Pre-authentication SQL injection (CVSS 9.8) enabling unauthenticated remote root command execution; actively exploited, added to CISA KEV on September 14, 2026. | 1, 2, 3 |
CVE-2026-76460 | Cisco Identity Services Engine (ISE) | Incorrect use of privileged APIs enabling exploitation; actively exploited, added to CISA KEV on September 16, 2026. | 1 |
CVE-2026-67277 | MikroTik RouterOS | Missing authentication for critical function enabling unauthenticated device takeover; actively exploited, added to CISA KEV on September 10, 2026. | 1 |
CVE-2026-86060 | MikroTik RouterOS | Argument delimiter injection enabling command injection; actively exploited, added to CISA KEV on September 10, 2026. | 1 |
CVE-2026-84411 | MikroTik RouterOS (web management service) | Integer underflow in HTTP request body handling enabling pre-auth remote code execution as root; no active exploitation reported as of September 29, 2026. | 1 |
CVE-2026-19490 | Citrix NetScaler ADC and Gateway | Authentication bypass on SAML IdP and Gateway/AAA-configured appliances; actively exploited, added to CISA KEV on September 9, 2026. | 1, 2 |
Endpoint software
| CVE | Affected product | Reported development | Sources |
|---|---|---|---|
CVE-2026-85046 | Google Chromium V8 | Type confusion vulnerability in V8 JavaScript engine exploited in the BlueMoon chain; added to CISA KEV on September 4, 2026. | 1, 2, 3, 4 |
CVE-2026-87491 | Google Chromium V8 | V8 sandbox escape via WebAssembly metadata corruption; exploited as part of the BlueMoon chain; added to CISA KEV on September 9, 2026. | 1, 2 |
CVE-2026-85880 | Microsoft Windows (ALPC and kernel) | Heap-based buffer overflow in ALPC enabling local privilege escalation; actively exploited in the wild, added to CISA KEV on September 8, 2026. | 1, 2, 3, 4 |
CVE-2026-81963 | Microsoft Windows (Update Stack) | Link following vulnerability enabling local privilege escalation; actively exploited in the wild, added to CISA KEV on September 8, 2026. | 1, 2, 3 |
CVE-2026-81578 | PaperCut NG/MF | Authentication bypass enabling unauthenticated access to privileged endpoints; actively exploited, chained with CVE-2026-82078 for RCE as SYSTEM. | 1, 2 |
CVE-2026-82078 | PaperCut NG/MF | Unsafe dynamic class-loading enabling arbitrary code execution as SYSTEM; actively exploited, added to CISA KEV on August 31, 2026. | 1, 2 |
CVE-2026-73570 | Zimbra Collaboration Suite (SNMP notification path) | Unauthenticated OS command injection via crafted SMTP request into SNMP trap processing; actively exploited, disclosed August 13, 2026. | 1 |
CVE-2026-87886 | Acronis Backup | Incorrect default permissions enabling unauthorized access or escalation; actively exploited, added to CISA KEV on September 16, 2026. | 1 |
CVE-2026-58704 | Google Pixel | Improper authorization vulnerability actively exploited in the wild; added to CISA KEV on September 16, 2026. | 1 |
CVE-2026-32996 | Veeam Agent for Microsoft Windows | Local privilege escalation via unbound elevated session UIDs in gRPC named pipe; public PoC released September 14, 2026. | 1 |
Cloud and developer tooling
| CVE | Affected product | Reported development | Sources |
|---|---|---|---|
CVE-2026-86218 | N-able N-central | Pre-auth static code injection (CVSS 10.0) enabling unauthenticated RCE; actively exploited, added to CISA KEV on September 8, 2026. | 1, 2, 3 |
CVE-2026-35273 | Oracle PeopleSoft (PSEMHUB) | Java deserialization flaw enabling unauthenticated RCE; exploited by ShinyHunters (UNC6240) via WAF bypass using URL-encoded path components. | 1 |
CVE-2026-87902 | WordPress Core | Remote file inclusion enabling execution of remote PHP files; actively exploited, added to CISA KEV on September 25, 2026. | 1 |
CVE-2026-65660 | Microsoft SharePoint Server | Code injection vulnerability with confirmed active exploitation; added to CISA KEV on September 25, 2026. | 1 |
CVE-2026-67394 | WebPros Plesk | Privilege escalation to root on Plesk servers prior to 18.0.79.9 and 18.0.80.5; no exploitation reported. | 1 |
CVE-2026-82329 | JFrog Artifactory | Improper authentication (CVSS 9.8) enabling unauthenticated admin token access; actively exploited, added to CISA KEV on September 2, 2026. | 1, 2 |
CVE-2026-84869 | ConnectWise ScreenConnect | Missing authorization (CVSS 9.9) enabling file transfer and execution via active remote sessions; actively exploited, added to CISA KEV on September 11, 2026. | 1, 2 |
CVE-2026-85706 | GitLab Community and Enterprise Editions | Path traversal enabling unauthenticated arbitrary file read; actively exploited, added to CISA KEV on September 11, 2026. | 1 |
CVE-2026-75650 | Adobe Commerce / Magento Open Source | Unauthenticated PHP code injection via template engine (CVSS 10.0); actively exploited since early September 2026, fixed in APSB26-146. | 1, 2 |
CVE-2026-62911 | Microsoft Exchange Server | Authentication bypass enabling mailbox takeover with basic privileges; public exploit code available, approximately 22,000 unpatched servers remain exposed. | 1 |
Industrial systems
| CVE | Affected product | Reported development | Sources |
|---|---|---|---|
CVE-2026-78012 | Pyramid Solutions NetStaX EtherNet/IP Stack | Critical stack-based buffer overflow (CVSS 9.8) in EtherNet/IP Stack prior to v5.6.1; no exploitation reported. | 1 |
CVE-2026-77393 | Inductive Automation Ignition | Blank default for Create Project Role(s) setting (CVSS 8.8) allowing any authenticated script executor to create projects; no exploitation reported. | 1 |
CVE-2025-10478 | Rockwell Automation 1756-ENBT Module | Network-exploitable denial-of-service flaw in ControlLogix EtherNet/IP bridge; all hardware versions affected, no patch available, vendor recommends hardware replacement. | 1 |
CVE-2026-18963 | Siemens Industrial Edge Management (Keycloak) | Critical password reset flaw (CVSS 9.1) in Keycloak reset-credentials flow enabling unauthenticated account takeover. | 1 |
CVE-2026-22755 | VIVOTEK Camera Firmware | Command injection (CVSS 10.0) in firmware modules across 37 camera models enabling unauthenticated remote root code execution; public PoC available, no active exploitation reported. | 1 |
Coverage
The source collection contained 268 articles.
Report classifications: Critical: 59; High: 162; Medium: 36; Low: 11.
Coverage by category: vulnerability: 115; malware: 61; general security news: 29; threat actor: 24; phishing/social engineering: 22; APT: 14; data breach: 3.
Counts describe collected reporting, not incidents or affected organizations.