Tahdid aktorlari ARTEX — avtonom penetratsion sinov tizimini — tarmoq hujumlarini tezlashtirish, imtiyozlarni oshirish va maxfiy ma'lumotlarni o'g'irlash uchun qurollantirmoqda.
Google Threat Intelligence Group va Mandiant tekshiruvlari taqdim etilgan hisobotda tasvirlanganidek, hujumchilar post-kompromiss operatsiyalarining katta qismini muvofiqlashtirilgan AI agentlariga topshirib, ijro ketma-ketliklarini soniyalarga qisqartirmoqda.
2026-yil avgust oxiridan kuzatilgan bu faoliyat moliyaviy motivatsiyali tovlamachilik guruhlari va shubhali kiberjosuslik operatorlarini qamrab oladi.
Bu Google'ning sentyabr oyidagi "So'rovdan avtonomiyaga" hisobotida hujjatlashtirilgan o'tishni — chatbot yordamidagi vazifalardan agentik ish oqimlariga (asboblarni bajarish, natijalarni talqin qilish va operatsion xatolarni bartaraf etish) — kengaytiradi.
Go tilida yozilgan va o'rnatilgan Next.js konsoli bilan ARTEX operator maqsadlarini mashina buyruqlariga olti ixtisoslashgan sub-agent orqali tarjima qiladi.
Tekshiruvchilar Active Directory razvedkasi, bulut ruxsatlarini aniqlash, hisob ma'lumotlarini yig'ish, imtiyozlarni oshirish va maxsus webshell joylashtirishni qamrab olgan qobiliyatlarni kuzatishdi.
Operatorlar ba'zan buzilgan hisob ma'lumotlarini taqdim etib, tizimga hujumni davom ettirishni buyurishdi. ARTEX keyin kirishni sinab ko'rdi, skriptlar yaratdi va xatolarga moslashdi.
Alohida ijro portlashlari taxminan ikki dan 25 soniyagacha davom etdi; bitta tekshiruvda tizim AWS ruxsatlarini tekshirayotganda 23 soniya ichida 27 hodisa qayd etildi.
Tezlik aniqlikka aylanmadi. Internetga ochiq Weaver OA serveriga qarshi ARTEX autentifikatsiyalangan ilova API'laridan SQL buyruqlarini berish uchun foydalandi va ma'lumotlar bazasi kirishini operatsion tizim ijrosiga aylantirishga urindi.
Uning iterativ yondashuvi 240 jurnal hodisasini yaratdi, jumladan raqobatlashuvchi sub-agentlar sabab bo'lishi mumkin bo'lgan xatolarni ham.
Tizim MySQL jurnalini suiiste'mol qilib, JSP webshell'larini vebga ochiq kataloglarga joylashtirishga urindi.
Kompilyatsiya xatolari ba'zi fayllar haqiqiy buyruq natijasini qaytarguncha takroriy qayta ko'rib chiqishlarga sabab bo'ldi — bu adaptiv ekspluatatsiya bilan birga sezilarli sud-tibbiy shovqinni ko'rsatadi.
GTIG ARTEX bilan bog'liq bitta kampaniyani UNC7228 sifatida kuzatadi, faoliyat Microsoft'ning Xitoy bilan bog'liq Storm-0940 belgilanishi bilan kesishadi.
Taqdim etilgan natijalar AQSh va Yevropadagi dasturiy ta'minot, sanoat avtomatlashtirish, yarimo'tkazgich va telekommunikatsiya tashkilotlarini qamrab olgan buzilishlarni tasvirlaydi.
Operatorlar odatda dastlab buzilgan hisobdan chuqur tizim kirishigacha taxminan to'rt soatda yetishdi, bitta holatda bu atigi 15 daqiqa davom etdi.
Wiz'ning oktyabr oyidagi tadqiqoti IT va dasturiy ta'minot sotuvchilari, jumladan Okta, Microsoft Entra ID, AWS Cognito va Microsoft 365 integratsiyalariga qarshi kesishuvchi faoliyatni mustaqil ravishda hujjatlashtirdi.
Hujum ketma-ketliklari ilova xaritalash, autentifikatsiya sinovi, parol purkash, hisobni tiklash suiiste'moli va o'zaro ijarachi kirish sinovini birlashtirdi.
Wiz agentlar yuklab olingan manba xaritalaridan API yo'llarini ajratib olganini, keyin esa bir necha daqiqa o'tib o'sha yo'llarga so'rov yuborganini kuzatdi.
To'siq bo'lganda, faoliyat nuqtalarni yoki infratuzilmani o'zgartirdi. Ba'zi sessiyalar 24–30 soat davom etdi, garchi ko'pchilik individual urinishlar muvaffaqiyatsiz tugagan bo'lsa ham.
Mandiant kuzatuvlari ma'lumotlar bazasidan ma'lumot olish, ommaviy blob o'qish, keng katalog ro'yxatlash va bulut saqlashni potentsial eksfiltratsiya uchun tayyorlashni o'z ichiga oldi.
Tekshiruvchilar shuningdek, aylanuvchi webhook.site nuqtalari orqali uzatilgan buyruq natijalarini aniqlashdi.
Himoyachilar tez ruxsat tekshiruvlari, takroriy avtorizatsiya xatolari, kutilmagan ma'muriy hisoblar, webshell yaratish, noodatiy snapshot faoliyati va tashqi webhook trafikini korrelyatsiya qilishlari kerak.
Alohida fayl nomlari yoki umumiy foydalanuvchi agentlari yetarli atributsiya signallari emas; Wiz ham ko'plab e'lon qilingan foydalanuvchi-agent ko'rsatkichlari mustaqil ravishda qidirilganda shovqinli ekanligini ogohlantiradi.
Tavsiya etilgan nazoratlar izolyatsiya qilingan boshqaruv tekisliklari, qisqa muddatli hisob ma'lumotlari, eng kam imtiyoz, segmentatsiya va tezkor cheklashni ishga tushirishga qodir xulq-atvorni aniqlashni o'z ichiga oladi.
Bu choralar agentik asboblar tezroq ekspluatatsiya qiladigan tanish zaifliklarga qaratilgan.
ARTEX ishlab chiquvchisi 8-oktyabrda zararli suiiste'moldan so'ng tizimni yopish va texnik xizmatni to'xtatishni e'lon qildi. E'lon mavjud nusxalar yoki fork'larni yo'q qilmaydi.
Hozirgi dalillar bir nechta hujum bosqichlarida muvaffaqiyatli, shovqinli avtomatlashtirishni ko'rsatadi — universal, yashirin, to'liq avtonom buzilish emas.
Darhol vazifa — avtomatlashtirilgan tajriba ekspluatatsiya qilinadigan kirishni topishidan oldin himoya javob oynalarini qisqartirishdir.
Threat actors are weaponizing ARTEX, an autonomous penetration-testing framework, to accelerate network intrusions, escalate privileges, and steal sensitive information.
Google Threat Intelligence Group and Mandiant investigations described in the supplied report show attackers delegating substantial portions of post-compromise operations to coordinated AI agents, compressing execution sequences into seconds.
Observed since late August 2026, the activity spans financially motivated extortion groups and suspected cyberespionage operators.
It extends the transition documented in Google’s September report, “From Prompting to Autonomy,” from chatbot-assisted tasks toward agentic workflows that execute tools, interpret results, and troubleshoot operational failures.
Written in Go with an embedded Next.js console, ARTEX translates operator objectives into machine commands through six specialized sub-agents.
Investigators observed capabilities covering Active Directoryreconnaissance, cloud permission discovery, credential harvesting, privilege escalation, and custom webshell deployment.
Operators sometimes supplied compromised credentials and instructed the framework to continue the intrusion. ARTEX then tested access, generated scripts, and adapted to errors.
Individual execution bursts lasted approximately two to 25 seconds; one investigation recorded 27 events across 23 seconds while the framework probed AWS permissions.
Speed did not translate into precision. Against an internet-facing Weaver OA server, ARTEX used authenticated application APIs to issue SQL commands and attempted to convert database access into operating-system execution.
Its iterative approach generated 240 log events, including failures that may have been caused by competing sub-agents.
The framework attempted to abuse MySQL logging to place JSP webshells in web-accessible directories.
Compilation errors prompted repeated revisions until some files returned valid command output, illustrating adaptive exploitation alongside substantial forensic noise.
GTIG tracks, one ARTEX-associated campaign as UNC7228, with activity overlapping Microsoft’s China-linked Storm-0940 designation.
The supplied findings describe compromises affecting software, industrial automation, semiconductor, and telecommunications organizations across the United States and Europe.
Operators typically progressed from an initially compromised account to deep system access in approximately four hours, with one case taking just 15 minutes.
Wiz’s October research independently documented overlapping activity against IT and software vendors, including Okta, Microsoft Entra ID, AWS Cognito, and Microsoft 365 integrations.
Attack sequences combined application mapping, authentication probing, password spraying, account-recovery abuse, and cross-tenant access testing.
Wiz observed agents extracting API routes from downloaded source maps, then requesting those routes minutes later.
When blocked, activity shifted endpoints or infrastructure. Some sessions persisted for 24–30 hours, although most individual attempts failed.
Mandiant’s observations included database extraction, mass blob reads, extensive directory enumeration, and preparation of cloud storage for potential exfiltration.
Investigators also identified command output transmitted through rotating webhook.site endpoints.
Defenders should correlate rapid permission checks, repeated authorization failures, unexpected administrative accounts, webshell creation, unusual snapshot activity, and outbound webhook traffic.
Individual filenames or generic user agents are insufficient attribution signals; Wiz similarly cautions that many published user-agent indicators are noisy when hunted independently.
Recommended controls include isolated management planes, short-lived credentials, least privilege, segmentation, and behavioral detection capable of triggering rapid containment.
These measures address familiar weaknesses that agentic tools exploit at greater speed.
ARTEX’s developer announced its closure and discontinued maintenance on October 8 following malicious abuse. The announcement does not eliminate existing copies or forks.
Current evidence demonstrates successful, noisy automation across multiple intrusion stages not universal, stealthy, fully autonomous compromise.
The immediate challenge is shortening defensive response windows before automated experimentation finds exploitable access.
Stops Cyber threats before impact with 21 min faster MTTR. Integrate ANYRUN’s Sandbox in your SOC.
