FBI ARTEX Threat Overview: TTPs, IOCs, and Investigation Items for Multi-Agent Attacks
Basic Information
- Article Title: Threat Overview of ARTEX Autonomous Penetration Testing Framework
- Publisher: Federal Bureau of Investigation (FBI)
- Publication Date: 2026-11-13
- Source:
FBI FLASH-20261113-001 - Related Sources: BleepingComputer: FBI warns ARTEX AI hacking tool is powering global cyberattacks, CrowdStrike: Unknown Threat Actor Uses AI-Driven ARTEX to Target South Korean Finance, Microsoft: Storm-0940 uses ARTEX following ConnectWise supply chain compromise
- Related Malware:
WISP,vshell,AdaptixC2,BLOODALCHEMY - Related Threat Actors:
Storm-0940,Storm-3358, Unidentified multiple threat actors - Related CVEs: None specified
- Related Products:
ARTEX,ARTEX PRO,Microsoft Entra ID,Microsoft Graph,Azure Kudu/SCM,AWS IAM - Importance (author-assessed response priority): High
Executive Summary
The FBI has synthesized cases where multiple threat actors used ARTEX to automate reconnaissance, authentication and cloud privilege checks, vulnerability exploitation, persistence, lateral movement, and data staging, and published IOCs along with investigation and mitigation items.
Attack Flow
Typical Flow Observed Across Multiple Campaigns
The following represents an aggregation by the FBI of observations across multiple campaigns and does not imply that the exact same order or techniques were used in every incident.
The ARTEX agent performs WHOIS lookups, directory fuzzing, and vulnerability scans, and rapidly checks the effective permissions of cloud credentials using dry-runs or test writes.
It repeatedly explores reachable vectors such as public application vulnerabilities, weak credentials, leaked VPN configurations, trust relationships with external partner tenants, and unsigned SAML assertions.
On successful pathways, it attempts to establish footholds and persistence using web shells, the Kudu administrative API, open-source RATs, additional cloud credentials, SSH keys, or OAuth application credentials.
It enumerates Domain Controllers and cloud resources, attempting lateral movement via Pass-the-Hash / Pass-the-Ticket, WinRM, or Kerberos relay. FBI documentation also includes lateral movement attempts blocked by endpoint protection.
It searches mailboxes, cloud storage, and backend databases, staging data using volumes or scripts to prepare for exfiltration.
Attacker Positioning and Execution Locations
External attackers connect to public applications, partner and supplier portals, cloud management planes, Microsoft 365, and business APIs from the internet or compromised relays / proxies.
Following initial access, compromised cloud identities, application identities, Windows / Linux servers, and Domain Controllers can serve as operational jumping-off points.
Visibility for Victims and Administrators
Users: When app-only authentication or public application exploitation proceeds without user interaction, clear indicators such as MFA prompts may be absent. Incidents may only be discovered later through the unauthorized use of mailboxes, personal information, or business data.
Administrators: Activity can be observed as high-frequency sequential requests, dry-runs or test writes, unusual service principal sign-ins, added OAuth credentials, Kudu API executions, web shells, rapid mass enumeration, staging volume creation, or transmissions to external webhooks.
Success and Failure Conditions
Success Conditions
The presence of at least one reachable attack surface, such as an internet-facing application, partner portal, cloud management plane, or leaked credentials.
The ability of the attacker to operate external LLM APIs and an ARTEX instance while maintaining network paths to the target.
Account or application identities post-initial access possess the permissions required for subsequent operations such as enumeration, credential modification, data access, and compute startup.
Failure Conditions and Mitigations
Inventory public-facing assets and management interfaces, and restrict Kudu/SCM, hypervisors, backup platforms, and cloud management planes to trusted networks.
Separate user identities from workload identities and apply the principle of least privilege, enforcing MFA for user authentication and credential rotation as well as application credential change monitoring for workload identities.
Block iterative scanning and lateral movement using WAFs, EDR, rate limiting, and network segmentation, while feeding blocked attempts into follow-up investigations.
Impact Upon Success
Cross-domain access to identity and data across cloud, SaaS, and on-premises environments.
Persistence established via web shells, RATs, OAuth credentials, SSH keys, and cloud roles.
Data collection, staging, and exfiltration from mailboxes, cloud storage, databases, and local systems.
The potential to reach privilege acquisition or data theft before incident response can initiate, driven by the parallel processing of multiple targets in a short timeframe.
Observable Logs
- Email: Check for reads, searches, bulk access, and mass notifications / broadcasts from compromised mailboxes.
- Proxy / SWG / DNS: Check for communications to external LLM APIs, unknown LLM gateways and resellers, Tor / SOCKS / residential proxies,
Chisel/SoftEther, and rotating webhook services. LLM API traffic alone does not confirm an attack. - Endpoint / EDR: Check for shells spawned from web or database processes,
JSP/PHP/ASPXweb shells,WISP/vshell/AdaptixC2, PowerShell enumeration, Pass-the-Hash / Pass-the-Ticket attempts, and EDR coverage probing. - Identity / IdP: Check for service principal sign-ins, authentications from unknown autonomous systems, hosting providers, or regions, added OAuth application credentials, cloud role modifications, password resets, and evidence of Kerberoasting or DCSync.
- SaaS / Cloud: Check for Microsoft Graph activity, Entra audits, Azure ARM dry-runs, sub-second blob writes / deletes, Kudu API calls, IAM
PassRoleusage, instance profile modifications, and unexpected VM, volume, Elastic IP, or SSH key creation. - Network: Check for high-frequency sequential requests to login / search endpoints, short bursts transitioning from scans to exploits, C2, and data transfers, and communications with FBI-published IOCs. IP addresses represent historical infrastructure at the time of observation and must be evaluated alongside current telemetry.
Attack Success Determination
Confirmations in Public Information
Attack attempts observed (success unconfirmed): The FBI has aggregated extensive scanning, dry-runs, authentication and privilege checks, and lateral movement attempts across multiple campaigns. Documentation includes Pass-the-Hash attempts blocked by endpoint protection; therefore, attempt counts are not treated as successful compromises.
Malware execution or successful authentication confirmed: In the ConnectWise campaign, Microsoft reports that Storm-0940 attempted sign-ins to over 2,250 tenants using ARTEX, with successful authentications and subsequent operations observed in multiple environments. While the FBI summary states this "enabled access to over 2,250 downstream tenants," Microsoft's description refers to sign-in attempt counts, and successful authentication alone does not indicate data access or exfiltration. The figure of over 2,250 is not treated as the number of successful authentications or victims across all targets.
Data theft or session compromise confirmed: In the South Korean financial campaign, personal data exfiltration has been confirmed in related documentation. The figure of approximately 67,000 individuals reflects the scope reported by the respective sources; it does not necessarily mean that every reported case has been independently verified by the affected organizations.
Subsequent compromise confirmed: Web shells, RATs, added OAuth credentials, cloud role manipulations, and database / mailbox / storage access have been confirmed in multiple incidents. In the ConnectWise campaign, subsequent operations such as Graph reconnaissance, application credential modifications, and privilege changes were also observed across multiple environments. These findings aggregate multiple campaigns and do not represent a single attack chain executed by a single threat actor.
Internal Assessment Criteria
Do not treat scans, dry-runs, or authentication failures as successful compromises; instead, verify token issuance, command execution, credential changes, data reads, staging, and egress on a stage-by-stage basis.
Treat high-frequency automation and LLM API usage as auxiliary indicators, corroborating them with identity, endpoint, cloud audit, and network telemetry.
Do not infer a compromise from a single hit on historical IOCs alone; correlate timestamps, requests, authentication results, processes, and data access.
Investigation Playbook
Investigation Starting Points: Begin with high-frequency sequential requests, service principal sign-ins from unknown sources, Kudu / Graph / ARM API anomalies, web shells, unusual egress to LLM gateways, and FBI-published IOCs.
Initial Review: Establish a chronological timeline of target assets, external reachability, authentication methods, execution privileges, initial successful responses, and immediate subsequent identity and cloud changes.
Endpoint / Server Investigation: Preserve web / database child processes, shells / RATs, PowerShell artifacts, credential access attempts, new accounts / SSH keys, and staged files / volumes.
Identity / Cloud Investigation: Review service principal sign-ins, OAuth credentials, role assignments, Graph / ARM / Kudu / IAM activity, cloud storage reads, and unexpected compute instances.
Subsequent Operations: Track internal enumeration, LAPS / Kerberos / DCSync, WinRM, mailbox / database / blob access, and egress via webhooks or proxies.
Containment: Revoke compromised identities and application credentials, restrict public-facing surfaces and management interfaces, and isolate web shells / RATs / tunnels. Rotate keys, tokens, and passwords after preserving evidence.
Categorization: Differentiate between reconnaissance, initial access, successful authentication, command execution, persistence, lateral movement, data access, staging, and exfiltration.
Defense and Detection Ideas
Single Events: Treat credential additions to high-privilege applications, arbitrary command execution via Kudu, unexpected role assignments, and the creation of JSP / PHP / ASPX files in web roots as high-priority items.
Chronological Correlation: Correlate scans / dry-runs -> authentication or exploitation -> credential / role changes -> internal enumeration -> staging -> egress within a short time window.
Threat Hunting: Cross-search public assets, partner trust relationships, service principals, Kudu/SCM, ARM dry-runs, blob test writes, new 100GB volumes, and unknown webhooks or LLM proxies.
Log Limitations: Because ARTEX can operate on external infrastructure, tool or model names may not leave traces in the victim environment. The absence of visible LLM API traffic does not prove the absence of ARTEX.
Priority Mitigations: Prioritize reducing external attack surfaces, restricting management plane network access, applying least privilege to workload identities, pre-collecting API / Graph / cloud audit logs, detecting credential modifications, and implementing rate limiting and segmentation.
Facts, Inferences, and Hypotheses
Facts
The FBI reported that since at least August 2026, multiple threat actors have abused ARTEX across global financial, government, technology, defense, semiconductor, energy, healthcare, manufacturing, telecommunications, and legal sectors.
ARTEX is a Go-implemented multi-agent platform that does not embed its own model, instead integrating with external models such as Claude, GPT, DeepSeek, GLM, Grok, and Kimi.
Observations aggregated by the FBI include cloud credential privilege checks, exploitation of SAML, XXE, SSRF, SQL injection, and SSTI, web shell and RAT deployment, cloud credential additions, Kerberoasting / DCSync, mailbox / storage / database access, and the use of proxies and tunnels.
Some intrusions were reported to have progressed from targeting to privileged access in under three hours, though this duration does not apply to all campaigns.
The original repository was made private on October 8, 2026, but forks, archives, localized variations, and ARTEX PRO remain available.
The figure of over 1,600 public ARTEX consoles identified as of early November 2026 is based on researcher internet observations and does not represent a count of instances confirmed by the FBI to have been abused.
FBI IOCs include historical infrastructure associated with South Korea, ConnectWise, and Storm-3358, and analysts are advised to account for cloud and virtual infrastructure reallocations by corroborating with current telemetry.
Inferences
The danger of ARTEX lies less in any specific model itself than in its ability to parallelize iterations across multiple assets, identities, and cloud APIs, thereby compressing the defender's decision window.
Because detection based solely on tool names, user-agent strings, or model APIs can be evaded, detection strategies should focus on sequences of privilege checks, credential changes, resource creation, and data access.
Hypotheses
No additional hypotheses. Unverified items are documented in "Unknowns and Additional Investigations."
MITRE ATT&CK Mappings
T1595Active Scanning - high - Directory fuzzing and active vulnerability scans have been observed.T1190Exploit Public-Facing Application - high - Exploitation of XXE, SSRF, SQL injection, SSTI, and command injection has been aggregated.T1078Valid Accounts - high - Stolen VPN credentials, cloud identities, and application identities are utilized.T1505.003Server Software Component: Web Shell - high -JSP/PHPweb shells and deployment via MySQL log poisoning have been observed.T1098.001Account Manipulation: Additional Cloud Credentials - high - Credential additions t
Comments
No comments yet. Start the discussion.