Skip to main content
Mallory
← Back to News
ai-enabled-threat-activitystate-sponsored-espionageoffensive-tooling-releaseinitial-access-method

Threat Actors Use ARTEX AI Framework to Automate Intrusions Across Hundreds of Targets

Emerging Technology Risk · 5 sources · Updated 17h ago · First seen Nov 19, 2026

Threat Actors Use ARTEX AI Framework to Automate Intrusions Across Hundreds of Targets

Threat actors are using ARTEX, an open-source, AI-powered penetration-testing framework, to automate reconnaissance, vulnerability discovery and exploitation. CERT-EU warned that the tool enables faster, more scalable attacks and assessed that it was highly likely involved in activity against its constituency; CERT Bulgaria subsequently highlighted the threat. Reporting also describes adoption by criminals with limited technical experience. According to SafeBreach’s summary of CrowdStrike research, the China-nexus espionage actor VAPOR PANDA used ARTEX during an 11-day campaign to target more than 200 organizations across at least 14 countries, with post-exploitation activity confirmed at nine organizations. Reported intrusion techniques included SAML validation failures, cloud metadata access, Active Directory abuse and stale credentials, alongside deployment of the BLOODALCHEMY backdoor. ARTEX accelerated established techniques through parallel agents and automated changes in attack paths rather than introducing new exploits.

ARTEX relies on external language models, while its scope restrictions and safeguards against destructive actions lack hard technical enforcement. The original author removed the repository following misuse reports, but public forks remain available. Tool usage alone is insufficient for attribution: multiple actors use ARTEX, and an independent source-code and laboratory analysis found that many published indicators originate from underlying tools or model-generated behavior rather than the framework itself. One popular fork also removed identifying strings after their publication. Defenders should prioritize correlated attack behavior over standalone strings, TLS fingerprints or timing patterns. Recommended actions include reducing internet exposure, patching exploitable weaknesses, rotating stale credentials, enforcing SAML signature validation and IMDSv2, and monitoring rapid authentication and cloud activity with comprehensive endpoint coverage.

New surface, new adversariesTrack how attackers are adapting to this technology.
Track this risk
Event timeline

How this story unfolded

Events from the most recent confirmed update back to the earliest known activity.

18 events · Jul 2026 – Nov 5, 2026
  1. Reported Nov 11, 20269d ago

    Censys identifies 1,465 publicly exposed ARTEX consoles

    Censys identified 1,465 exposed consoles representing roughly 815 distinct-looking deployments after accounting for duplicated configurations. It found additional threat tooling on 155 ARTEX hosts, while cautioning that not all deployments were malicious and the number lacking authentication was uncertain.

    ARTEX AI Pentest Framework: How VAPOR PANDA Hit 200+ Orgs
  2. Reported Nov 11, 20269d ago

    VAPOR PANDA uses stale keys to access Greek government mailboxes and a VPN

    Attackers used apparently unrotated authentication keys from a 2022 mail-server directory export to forge an administrative token and read Greek government mailboxes. They then used a VPN configuration found in sent mail and a client key reportedly originating from a May 2024 email leak to reach the internal network.

    ARTEX AI Pentest Framework: How VAPOR PANDA Hit 200+ Orgs
  3. Reported Nov 11, 20269d ago

    A second pharmaceutical intrusion reaches root and privileged account impersonation

    Another pharmaceutical intrusion involved web shells and an unauthenticated orchestration service granting root access in a production container environment. Attackers also retrieved LAPS passwords and used resource-based constrained delegation to impersonate a privileged account.

    ARTEX AI Pentest Framework: How VAPOR PANDA Hit 200+ Orgs
  4. Reported Nov 11, 20269d ago

    VAPOR PANDA chains SAML and XML flaws in a pharmaceutical intrusion

    Attackers combined acceptance of an unsigned SAML assertion with an XML external entity flaw to access cloud instance metadata and obtain a role capable of executing commands on thousands of instances. The intrusion subsequently recovered an enterprise administrator account without requiring a zero-day vulnerability.

    ARTEX AI Pentest Framework: How VAPOR PANDA Hit 200+ Orgs
  5. Reported Nov 8, 202612d ago

    Source-code and laboratory analysis challenges the reliability of ARTEX indicators

    A researcher evaluated 62 published indicators and surveyed 73 ARTEX code bases, finding that most target-side indicators originated from models or ordinary tools rather than the framework. Laboratory tests also showed that optional traffic recording altered TLS fingerprints and that pauses between activity bursts largely reflected model response times, limiting their value as definitive ARTEX identifiers.

    ARTEX: reading the source behind the IOCs · andpalmier's blog
  6. Nov 5, 202615d ago

    CERT-EU warns of widespread ARTEX abuse and likely targeting of its constituency

    CERT-EU warned that multiple threat actors were using ARTEX to automate attacks against vulnerabilities and misconfigurations, and assessed that similar activity against its constituency highly likely involved the framework. It recommended reducing internet exposure, prioritizing patching, strengthening monitoring, and addressing minor weaknesses that automated agents could chain together.

    Злоупотреба с автономната AI платформа ARTEX за извършване на кибератаки - CERT Bulgaria
  7. Nov 4, 202616d ago

    CERT-In warns financial and cryptocurrency organizations about Crafty Falcon

    India's CERT-In warned that Crafty Falcon used ARTEX alongside PentAGI, Cairn, and CyberStrikeAI. The warning provided further evidence that ARTEX usage alone does not uniquely identify a threat actor.

    ARTEX AI Pentest Framework: How VAPOR PANDA Hit 200+ Orgs
  8. Nov 3, 202617d ago

    RuoJi6/ARTEX removes identifying public-response strings

    Four days after Unit 42 published the authentication-realm indicator, RuoJi6/ARTEX removed identifying public-response strings. Changes included replacing the authentication realm with "Restricted," moving version information behind authentication, and removing identifying health-response fields.

    ARTEX: reading the source behind the IOCs · andpalmier's blog
  9. Oct 30, 202621d ago

    Unit 42 publishes ARTEX's authentication-realm indicator

    Unit 42 published the ARTEX Basic authentication realm as an indicator for identifying exposed framework infrastructure.

    ARTEX: reading the source behind the IOCs · andpalmier's blog
  10. Oct 30, 202621d ago

    Google and Mandiant report offensive AI frameworks in real intrusions

    Google Threat Intelligence and Mandiant reported that threat actors were using ARTEX, PentAGI, Hexstrike, and CAIRN in real intrusions.

    ARTEX: reading the source behind the IOCs · andpalmier's blog
  11. Oct 29, 202622d ago

    CrowdStrike attributes the multinational ARTEX campaign to VAPOR PANDA

    CrowdStrike attributed the campaign to the China-nexus espionage actor VAPOR PANDA, linking its ARTEX controller to actor infrastructure. This was the actor's first observed use of AI tooling and expanded its observed targeting into pharmaceutical, semiconductor, nuclear, and defense organizations.

    ARTEX AI Pentest Framework: How VAPOR PANDA Hit 200+ Orgs
  12. Oct 21, 20261mo ago

    Microsoft discloses Storm-0940's ARTEX-enabled ConnectWise compromise

    Microsoft reported that Storm-0940 used ARTEX in a ConnectWise supply-chain compromise followed by sign-in attempts against more than 2,250 Microsoft 365 tenants. Some intrusions reached privileged access in under three hours.

    ARTEX AI Pentest Framework: How VAPOR PANDA Hit 200+ Orgs
  13. Oct 17, 20261mo ago

    VAPOR PANDA begins an ARTEX campaign targeting more than 200 organizations

    An 11-day campaign beginning October 17 used one ARTEX controller to target more than 200 organizations across at least 14 countries. CrowdStrike subsequently confirmed post-exploitation activity involving BLOODALCHEMY at nine organizations, including a Middle Eastern interior ministry and a Latin American foreign ministry.

    ARTEX AI Pentest Framework: How VAPOR PANDA Hit 200+ Orgs
  14. Oct 10, 20261mo ago

    Storm-0940 password spray fails against accounts at eight organizations

    A password-spraying operation described in Microsoft's Storm-0940 reporting ran October 10–14 against 327 accounts at eight organizations. It produced no successful authentication despite peaking at 2,900 failed sign-ins per minute.

    ARTEX AI Pentest Framework: How VAPOR PANDA Hit 200+ Orgs
  15. Oct 9, 20261mo ago

    RuoJi6/ARTEX adds optional Basic authentication with an identifying realm

    The RuoJi6/ARTEX fork added an optional HTTP Basic authentication gate using the realm "ARTEX." Upstream ARTEX did not provide HTTP Basic authentication.

    ARTEX: reading the source behind the IOCs · andpalmier's blog
  16. Oct 7, 20261mo ago

    CrowdStrike reports ARTEX attacks against South Korean financial organizations

    CrowdStrike reported that an unattributed Chinese-speaking, financially motivated operator used ARTEX against South Korean financial organizations, assessing attribution at moderate confidence. Exposed servers revealed session histories, configurations, and memory files, but victim identities and counts remained unsettled.

    ARTEX AI Pentest Framework: How VAPOR PANDA Hit 200+ Orgs
  17. Oct 2026

    ARTEX developer removes the original repository following misuse reports

    The original developer made the ARTEX repository unavailable following reports of misuse and denied involvement in attacks. Independent forks remained publicly available.

    CERT-EU - From Penetration Testing to Cyber Threat: Beware of ARTEX
  18. Jul 2026

    Autumn-27 releases ARTEX as an open-source penetration-testing framework

    A developer using the name Autumn-27 published the Go-based ARTEX framework on GitHub for authorized security testing. The framework uses AI planner and worker agents to automate reconnaissance and attack workflows.

    ARTEX AI Pentest Framework: How VAPOR PANDA Hit 200+ Orgs
Linked entities

Related entities

Vulnerabilities, threat actors, malware, products, organizations, and breaches Mallory has linked to this story.

Open in app →

Threat actors

1 linked

Malware

1 linked
Do you run this?Follow how adversaries are adapting to this technology, and where it touches your stack today.
Track This Risk
Mallory platform

See the full picture, correlated to your attack surface.

This page covers what’s public. Mallory adds the parts that aren’t: which of your assets are affected, which threat actors are using it right now, and what to deploy next.

Exposure mappingMap indicators from this story to your assets and identify affected systems in minutes.
Threat actor evidenceEvery observed campaign, victim, and pivot linked to actors named in this story.
Associated malwareMalware, exploits, and IOCs connected to the activity described here.
Detection signaturesYARA, Sigma, and Snort rules deployed to your SIEM as soon as they’re published.
Scheduled alertsGet matching new stories delivered to your team as they break, not the next morning.
AI threadsAsk questions about this story and take action on the answers.