Vapor Panda Expands ARTEX Attacks
CrowdStrike confirmed post-breach activity in nine organizations and new ARTEX signs, with BloodAlchemy and intrusion vectors detailed.
CrowdStrike confirmed post-breach activity in nine affected organizations, including the installation of the BloodAlchemy backdoor, and observed password spraying against at least 12 others. The campaign also hit U.S. semiconductor, defense, think tank and telecommunications organizations.
Update November 10, 2026: RH-ISAC released concrete indicators to look for artifacts linked to ARTEX, while Pixee detailed new vectors observed in the campaign. SOCRadar also expanded BLOODALCHEMY's reach to a Middle Eastern interior ministry and a Latin American foreign ministry, without identifying the countries.
CrowdStrike confirmed post-intrusion activity in nine affected organizations, including the installation of the BloodAlchemy backdoor, and observed password spraying against at least 12 others. RH-ISAC said the campaign also reached U.S. semiconductor, defense, think tank and telecommunications organizations.
CrowdStrike said a single ARTEX controller was used for 11 days against more than 200 organizations in at least 14 countries in a campaign attributed to Vapor Panda. The technical analysis also points to ARTEX being used alongside DeepSeek and another commercial model, and describes parallel intrusions carried out against high-value targets.
What is known about the scope of the campaign?
The operation had a defined scale in both time and volume, with activity sustained for 11 days and reaching more than 200 organizations, according to Cybersecurity Dive and CrowdStrike's analysis. The technical source adds that the pattern was not isolated, but part of parallel execution against high-value targets.
That deployment matters because it points to a coordinated, repeated campaign rather than a one-off incident. The combination of ARTEX with DeepSeek and another commercial model was part of the technical framework CrowdStrike described for automating and scaling the activity.
What indicators did ARTEX leave behind to track the campaign?
RH-ISAC said CrowdStrike observed Vapor Panda using ARTEX together with DeepSeek and another commercial model, and recommended looking for artx or artex commands, cloud roles with the artx prefix, and the directory C:\ProgramData\artx5v. Those traces can help identify possible operational fingerprints in compromised environments.
The same organization said those elements appear in the context of a campaign that combined automation with AI agents. In practical terms, the indicators point to the console and terminal, as well as cloud configurations and local paths within affected systems.
What happened in Latin America?
According to ITBrief Asia, Vapor Panda carried out intrusions against foreign ministries in two Latin American countries and likely compromised a third. The available source does not identify which countries were affected, so it is not possible to narrow the attribution further.
SOCRadar added that CrowdStrike documented BLOODALCHEMY in a Middle Eastern interior ministry and a Latin American foreign ministry, though it did not identify the countries. That broadens the regional reference without resolving the geographic attribution.
A separate secondary report also said at least nine organizations were compromised during the operation, in addition to the targets that were probed. That figure, however, does not include a verifiable breakdown by country or sector, so it does not show how far each front of the campaign reached.
What technical vectors did the investigation observe?
Pixee said the campaign included an unsigned SAML assertion combined with XXE, an unauthenticated orchestration service with root access in a container, keys from a 2022 directory export, and VPN configurations stored in sent email messages. Those vectors show the intrusion did not rely on a single technique.
The same report said CrowdStrike confirmed post-exploitation activity in nine organizations and password spraying, or credential scanning, against at least 12 others. That technical picture matches an operation that combined initial access, persistence and credential hunting.
What can be stated with technical confidence?
The strongest element in the case is CrowdStrike's confirmation that ARTEX was used as a single controller for 11 days, alongside AI models to sustain the campaign. That combination points to an operation designed to increase speed and volume, but the public material leaves gaps about the countries, the agencies and the exact extent of the compromises.
RH-ISAC added that the campaign reached U.S. semiconductor, defense and think tank organizations, as well as telecommunications providers. That coverage extends the map beyond Latin America and reinforces that the operation had targets on several fronts.
Microsoft described ARTEX as an open-source penetration testing framework capable of working with different AI models, a detail that helps explain why the tool could adapt to different workloads. Industrial Cyber, meanwhile, reported that the campaign combined ARTEX with the China-developed DeepSeek model and another commercial model, and described it as an architecture of planner and worker agents for reconnaissance, credential theft, lateral movement and web shell deployment.
Industrial Cyber also reported that the ARTEX controller directed exploitation activity against entities in at least 14 countries for 11 days and that the framework coordinated reconnaissance, credential theft, lateral movement and web shell deployment. That characterization matches the intrusion pattern described by CrowdStrike and helps place the campaign within a sustained exploitation sequence.
BloodAlchemy was first documented publicly in October 2023, according to RH-ISAC. Regionally, ESET added that FamousSparrow maintained monitoring activity in South America between April and September 2026, although the available report does not specify countries, organizations or concrete intrusions. Based on the sources provided, the regional map is still defined by partial attributions and incomplete detail.
Sources
- VAPOR PANDA hackers use multi-agent AI to target government, pharmaceutical, semiconductor, defense organizationsindustrialcyber.co· Industrial Cyber
- Security Firms Identify Additional Global Threat Campaigns Leveraging ARTEX AI Pentest Platformrhisac.org· RH-ISAC
- Hackers let an AI run break-ins at education organizationsintelfusions.com· IntelFusions
- ESET APT Activity Report Q2–Q3 2026: China, Iran, North Korea ...aviatrix.ai· Aviatrix
- CrowdStrike uncovers China-linked AI cyber campaignitbrief.asia· ITBrief Asia
- Chinese hackers use ARTEX AI in automated attacks against 200 orgsbleepingcomputer.com· BleepingComputer
- AI agents scale and speed up Chinese-linked attackshelpnetsecurity.com· Help Net Security
- VAPOR PANDA Uses AI-Driven ARTEX to Scale a Global Espionage Campaigncrowdstrike.com· CrowdStrike
- Agentic pen-testing tool employed in multiple espionage campaignscybersecuritydive.com· Cybersecurity Dive
- China-linked hackers use AI tools to target 200 organizations, compromising ninecisovoice.com· CISO Voice
- Chinese Hackers Use ARTEX AI to Target 200 Organizations Across 14 Countriesotecosystem.com· Oteco System
- Hackers Ran an AI Attack Framework Against 200 Organizations in 11 Dayspixee.ai· Pixee
- ESET APT Activity Report Q2 2026–Q3 2026esetngblog.com· ESET
- Boletín Semanal de Ciberseguridad, 6 de noviembretelefonicatech.com· Telefónica TechUnverified URL
- ARTEX AI Pentest Tool: What Security Teams Need to Knowsocradar.io· SOCRadar



