CrowdStrike reported on Oct 29 that the China-nexus espionage group VAPOR PANDA, starting Oct 17, ran the open-source agentic intrusion framework ARTEX with DeepSeek plus a second, unconfirmed commercial LLM, hitting more than 200 organizations in at least 14 countries from a single controller over 11 days. Post-exploitation was confirmed at nine organizations, where the BLOODALCHEMY backdoor was deployed. Planning agents used about 1.4 billion input tokens in a single intrusion, and recovered transcripts show a human operator steering the agents in Chinese at key decision points.
Key Takeaways
- ✓Over 11 days from Oct 17, one ARTEX controller targeted 200+ organizations in at least 14 countries, including several foreign ministries, 19 US pharma, semiconductor, defense, nuclear and finance organizations, and telecoms in the US, Belgium and the Philippines (CrowdStrike report)
- ✓Post-exploitation was confirmed at 9 organizations, which received the BLOODALCHEMY backdoor; at least 12 more saw password spraying and credential scanning
- ✓DeepSeek was the main model, plus a subscription commercial LLM CrowdStrike could not confirm; planning agents used about 1.4 billion input tokens in one intrusion
- ✓Recovered transcripts show an operator issuing Chinese instructions tagged as human instructions, including an order to find Windows hosts without the Falcon sensor
- ✓CrowdStrike says this is separate from the Storm-0940 ARTEX activity reported by Microsoft, and judges China-nexus groups will highly likely keep adopting such agent frameworks over the next 12 months
Key Decision Metrics at a Glance
Turn your technical choice into a development budget
Compare 40 dev plans & simulate token costs vs $20/mo subscriptions
Project Links & Resources
Direct AccessIn-Depth Technical Analysis
Background and the problem
ARTEX, an open-source multi-agent pentest framework built by Chinese developers, keeps turning up in real attacks. On Oct 10 we covered CrowdStrike finding an attacker who used ARTEX with Claude Code against South Korean financial firms, and Microsoft later reported Storm-0940 using ARTEX in a supply-chain compromise. This new CrowdStrike report is a step up: the operator is a state-linked espionage group, VAPOR PANDA, and the scale went from single intrusions to more than 200 targets at once. For defenders, the problem is that agents let a small team cover an attack surface that used to need many people.
How it works
Per the report, ARTEX runs a main agent over planner and worker agents that share a graph of facts, findings and goals, moving each job from initial access to host reconnaissance and lateral movement. When one path fails it switches to another, and it tries web apps, cloud services, containers, Active Directory, mail and VPN in parallel. A human operator stayed in the loop, giving Chinese instructions at decision points such as writing web shells, pulling domain admin passwords and listing hosts without the Falcon sensor. One case chained an unsigned SAML assertion with an XXE flaw to read cloud instance metadata and obtain a role that could send admin commands to thousands of instances.
Benchmarks
This is a threat report, so there are no model benchmarks. All figures are CrowdStrike's: 200+ targets, at least 14 countries, 11 days, 9 confirmed compromises, 12 more sprayed, and about 1.4 billion input tokens in one intrusion. CrowdStrike could not identify the second commercial LLM, and we do not guess.
What developers and defenders can do
CrowdStrike's recommendations work as a checklist: rotate signing keys and service credentials after any suspected exposure, including old backups; enforce SAML signature checks, require IMDSv2 and limit cloud roles' command-execution rights; hunt for artx or artex strings in command lines, artx-prefixed cloud roles and the C:\ProgramData\artx5v\ folder; and patch internet-facing appliances. The report ends with BLOODALCHEMY C2 domains, IPs and file hashes you can load into detection tools.
Benchmark side-by-side against alternatives, or calculate monthly token cost vs subscription break-even.
Discussion & Comments
0Sign in to join the discussion
Connect with AI developers to exchange benchmark insights.