Silicon Scoop · Technology
ARTEX AI Tool Used to Target 200+ Orgs, CrowdStrike Says
Craig Major·November 17, 2026
ARTEX AI is used in two separate campaigns, Microsoft and CrowdStrike report. Read the FBI ARTEX advisory takeaways and what founders should do now.
Estimated reading time: 10 minutes
Key Takeaways
- CrowdStrike said Vapor Panda used ARTEX AI in a campaign that began October 17 and had targeted 200+ organizations in at least 14 countries by November 4. At least nine were compromised.
- The FBI said some ARTEX intrusions went from first targeting to privileged access in under three hours.
- Microsoft tied Storm-0940 to ARTEX use in the ConnectWise Cloud Backup incident. It said the actor tried to authenticate to more than 2,250 downstream Microsoft 365 tenants.
- The original developer pulled ARTEX from GitHub on October 8, but forks, archives and a paid ARTEX PRO edition remain. Researchers had found 1,600+ exposed consoles by early November.
- Most evidence comes from security vendors reporting on their own findings. Treat the numbers as company-reported, not independently audited.
Table of contents
- What the FBI ARTEX Advisory Says About Speed and Skill
- How ARTEX AI Works, and Why Deleting It Did Not Stop It
- Storm-0940 ARTEX Abuse in the ConnectWise Cloud Backup Breach
- Vapor Panda ARTEX Campaign: 200+ Targets, 9 Compromises
- ARTEX AI in South Korea: Bank Breaches and 8.2 Million Records
- What the Vendor Reports Agree On, and What Is Still Unclear
- What This Means for Startups and Small Businesses
- Frequently Asked Questions
CrowdStrike said a China-linked group it calls Vapor Panda used ARTEX AI in a campaign that began on October 17. By November 4, the campaign had targeted more than 200 organizations across at least 14 countries. CrowdStrike said at least nine of those organizations were compromised.
ARTEX is an open-source framework that uses AI agents to run much of an attack. The FBI says it lets people with little technical skill run complex, multi-target operations.
For founders and operators, the point is speed and cost. Agentic AI tools (software that takes a goal and carries out many steps on its own) can shrink the time defenders have to react. They push up the cost of cloud and identity security. Small teams with fewer controls are the most exposed.
This article covers what the FBI said, what Microsoft and CrowdStrike reported, and what to do about it.
What the FBI ARTEX Advisory Says About Speed and Skill
The FBI issued a FLASH advisory (a short alert for network defenders) about ARTEX. BleepingComputer's report on the advisory says it was released on a Friday and runs 13 pages. The briefing's timeline dates the warning to November 14, so this article does not give an exact release day.
The FBI said multiple threat actors have used the open-source framework since at least August 2026. Targets include the financial, government, technology, defense and healthcare sectors.
The under-three-hours claim
The advisory states:
> "ARTEX significantly reduces the time required to chain attack techniques, with some intrusions progressing from initial targeting to privileged access in under three hours."
"Privileged access" means the attacker gets admin-level control. That is the stage where real damage starts.
Why skill barriers matter
The FBI also said the framework "enables threat actors with minimal technical expertise to conduct complex, multi-target operations simultaneously."
In plain terms, a smaller or less skilled group can now do work that once took a larger, more skilled team. The same operator can also attack many organizations at once.
What the advisory includes
The advisory lists indicators of compromise. These are clues, such as IP addresses, command-and-control servers and suspicious network activity, that show a system may have been attacked.
The FBI cautioned that listed IP addresses may have changed since the attacks happened. Defenders should not rely on them alone.
These details come from the news report on the advisory, not from the advisory itself. The research for this article did not include a link to the FBI document.
Source: BleepingComputer report on the FBI ARTEX advisory
How ARTEX AI Works, and Why Deleting It Did Not Stop It
ARTEX is an autonomous penetration-testing framework written in Go, a programming language. Penetration testing means hunting for weak spots in a system, which is normally done by defenders. ARTEX uses AI agents to do that work with little human input.
According to the FBI, the agents can:
- Scout a target (reconnaissance)
- Find vulnerabilities
- Exploit systems
- Raise their own privileges
- Steal data
It borrows its AI brains
ARTEX does not ship with its own AI model. It connects to outside large language models (LLMs), which are the AI systems behind chatbots. The FBI named:
- Anthropic Claude
- OpenAI GPT
- DeepSeek
- GLM
- Grok
- Kimi
An attacker gives ARTEX goals and tasks. The framework turns them into commands that specialized agents run. That is how one operator can work against many targets at once.
Where it finds the cracks
The FBI said ARTEX is especially good at finding weak spots in systems people forget about:
- Partner portals
- Employee applications
- Legacy services
- Third-party integrations
Small companies tend to have plenty of these. A forgotten partner login page can matter as much as the main website.
The FBI noted that automated attacks can create lots of noisy traffic that trips alerts. But the speed may leave defenders little time to respond before a breach.
Pulled from GitHub, still in circulation
A developer known as Autumn-27 first published ARTEX on GitHub in July 2026. The developer removed the repository on October 8, after reports of abuse.
The FBI said ARTEX is still available through:
- Community forks (copies other people made)
- Archived copies
- Localized versions
- A commercial edition called ARTEX PRO
As of early November, researchers had found more than 1,600 publicly exposed ARTEX consoles. The real number of deployments was believed to be significantly higher.
Source: BleepingComputer report on the FBI ARTEX advisory
Spend your hours on growth, not on repetitive busywork
Security is one demand on a small team's time. Manual, repetitive tasks are another, and they pile up fast. Flowgrammer.ca builds business automation that takes routine work off your plate, so you can scale without hiring for every new task.
Storm-0940 ARTEX Abuse in the ConnectWise Cloud Backup Breach
Microsoft published its investigation on October 21, 2026. It said Storm-0940, a China-state-aligned actor, used ARTEX to automate attacks against hundreds of organizations globally. Microsoft said the actor began using ARTEX as early as September 2026.
BlackBerry's briefing repeats Microsoft's attribution, including the "since at least September" timing. It does not appear to be separate confirmation.
What happened in the supply-chain attack
A supply-chain attack hits a vendor so the attacker can reach the vendor's customers. Here, the vendor service was ConnectWise Cloud Backup. BleepingComputer's report says the service was formerly called SkyKick Cloud Backup.
Microsoft said ConnectWise published an initial advisory on October 16, 2026. Microsoft said the attackers abused compromised application identities. These are the digital credentials software uses to log in on its own.
With them, attackers could sign in to customer Microsoft 365 environments. They used permissions the customers had already granted to those apps. No stolen user password was needed for that step.
The numbers, and what each one means
These figures measure different things, so they should not be mixed up:
- 2,250+: the number of downstream Microsoft 365 tenants against which the attackers attempted to authenticate, according to Microsoft. This counts attempts, not successful breaches.
- 162: the number of compromised tenants Huntress reported identifying, as relayed by BleepingComputer. The report says "at least" 162.
- Hundreds: the number of organizations Microsoft said were affected globally.
According to BleepingComputer's summary of Microsoft's findings, ARTEX sped up three things:
- Reconnaissance
- Password spraying (trying common passwords across many accounts)
- Abuse of cloud-application permissions
Some intrusions went on to change credentials and try to set up persistent access, so the attackers could get back in later.
Secondary coverage
CERT-EU's November 6 cyber brief summarizes Microsoft's assessment that Storm-0940 used ARTEX to speed up intrusions involving ConnectWise cloud applications. It summarizes Microsoft's work rather than adding independent findings.
Source: Microsoft Security Blog, October 21; BlackBerry briefing, November 6; CERT-EU cyber brief, November 6; BleepingComputer report on the FBI ARTEX advisory
Vapor Panda ARTEX Campaign: 200+ Targets, 9 Compromises
CrowdStrike published its Vapor Panda report on November 16, 2026. It describes a campaign it says began on October 17. By November 4, CrowdStrike said, the group had targeted more than 200 organizations in at least 14 countries. It said at least nine organizations were compromised.
The gap between targeted and compromised matters. Targeting means the tool went after an organization. Compromise means the attacker got in. Based on CrowdStrike's figures, nine compromises is a small share of 200 targets. Even so, nine is the minimum, and CrowdStrike's report words it as "at least."
Date check
PCrisk, a secondary security-news site, reports that CrowdStrike OverWatch (CrowdStrike's threat-hunting team) spotted the campaign on October 17. CrowdStrike's own summary lists October 17 as the campaign start. The research does not settle whether that was the first attacker activity or the first detection.
This article therefore does not claim a precise campaign length. The sources say the campaign began October 17 and had reached 200+ organizations by November 4. They do not say the 200+ figure came in a shorter window.
What the AI agents did
BleepingComputer's summary of CrowdStrike's findings says the agents:
- Performed reconnaissance
- Exploited vulnerabilities
- Deployed web shells (small scripts that give attackers remote access to a web server)
- Stole credentials
- Pursued different attack paths when initial attempts failed
That last point is the key difference from older automated attacks. A basic script gives up when a door is locked. An agent can try another door, which makes it harder to stop with a single control.
Why a company-reported number still counts
These are CrowdStrike's findings. They are not audited by outside parties. Still, the report comes from a major incident-response firm that says it saw the activity directly.
Vapor Panda and Storm-0940 are different named actors. The campaigns are described in separate reports, with different start points. The sources do not say they are linked.
Source: CrowdStrike Vapor Panda report, November 16; PCrisk report, November 3; BleepingComputer report on the FBI ARTEX advisory
ARTEX AI in South Korea: Bank Breaches and 8.2 Million Records
The FBI advisory also pointed to cases in South Korea, according to BleepingComputer's report. These are described as separate from the Storm-0940 and Vapor Panda campaigns.
Banks hit with ARTEX and Claude Code
In one case, a Chinese-speaking threat actor combined ARTEX with Claude Code agents. Claude Code is an AI coding assistant. The attacker used these tools against multiple South Korean banks, including:
- Shinhan Bank
- KB Kookmin Bank
- Hana Bank
The attacks exposed personal information on about 67,000 people.
CrowdStrike researchers found the attacker's exposed infrastructure. It held ARTEX configuration files, Claude Code session histories and other records documenting the intrusions. In effect, the attacker left a log of the operation online.
Seven arrests
In a separate investigation, South Korean police arrested seven suspects. They were linked to attacks on more than 90 organizations. ARTEX was reportedly used in most of the intrusions.
Authorities said the gang stole about 8.2 million data records. Victims included:
- Hospitals
- Educational institutions
- Businesses
Note the word "reportedly." The report says ARTEX was used in most intrusions, not all. The figure of 90 organizations covers all the gang's attacks, not only those involving ARTEX.
Why these cases matter to small firms
Hospitals, schools and ordinary businesses were among the victims. These are not only big-name targets. Attackers using automation can afford to go after smaller organizations because each target costs them little effort.
Source: BleepingComputer report on the FBI ARTEX advisory
What the Vendor Reports Agree On, and What Is Still Unclear
Microsoft, CrowdStrike, Okta, CERT-EU and BlackBerry all describe ARTEX as speeding up reconnaissance, password spraying, abuse of cloud-app permissions and follow-on cloud compromise. These are security vendors and agencies reporting on their own work.
No independent academic or third-party analysis was available in the research for this article. This is therefore not an expert consensus, only a pattern across company reports.
Several of these sources also depend on one another. CERT-EU summarized Microsoft's assessment. BlackBerry repeated Microsoft's attribution. They add reach, not independent proof.
What Okta added
Okta Threat Intelligence published an analysis on November 5. It said multiple actors likely used agentic frameworks, likely ARTEX, for broad identity-centered reconnaissance and compromise attempts. It also said a subset of that activity overlaps with Microsoft's Storm-0940 attribution.
"Likely" is Okta's word, and it signals an estimate, not a proven finding. Okta's wording suggests several actors are using the same tool, rather than one coordinated operation.
Where the disagreement sits
The reports do not really dispute that ARTEX is being abused. The open questions are how widely and by whom:
- Microsoft stresses China-state-aligned Storm-0940 and supply-chain abuse.
- CrowdStrike describes Vapor Panda, a Chinese cyberespionage group, in a separate campaign.
- The FBI describes multiple actors, plus other cases such as the South Korean banks.
- Okta says the activity likely spans several actors.
Reading these together, multiple groups appear to have adopted the same framework. That is a reasonable reading, not a confirmed fact. No source in the research lists every actor or every victim.
A caution on timelines
The reports cover different events on different dates. Storm-0940's ARTEX use dates to September, the Vapor Panda campaign to October 17, and the FBI's "since at least August" to multiple actors. These are separate milestones. Do not treat them as a single timeline or compare them to infer delays.
Source: Microsoft Security Blog, October 21; CrowdStrike Vapor Panda report, November 16; Okta Threat Intelligence, November 5; CERT-EU cyber brief, November 6; BlackBerry briefing, November 6
What This Means for Startups and Small Businesses
Agentic attack tools lower the cost of attacking you. They can also shrink the time you have to respond. A small team with a few admins and a pile of third-party apps is exactly the kind of target the FBI described.
Start with identity and app permissions
The ConnectWise case turned on application identities and permissions customers had already granted. Check these first:
- List every third-party app connected to your Microsoft 365 or other cloud tenant.
- Remove apps you no longer use.
- Review what permissions each app holds, and cut back anything broader than it needs.
Follow the FBI's guidance
According to BleepingComputer's report, the FBI urged organizations to:
- Secure internet-facing applications
- Restrict access to cloud-administration interfaces
- Review application permissions and authentication logs
- Add rate limits on sensitive APIs
- Monitor for unusual bursts of automated requests
- Monitor for suspicious cloud activity
- Monitor for unexpected credential or privilege changes
- Monitor for large data transfers
Rate limiting caps how many requests a user or system can make in a given time. That blunts automated password guessing.
Check the forgotten corners
The FBI said ARTEX is good at finding partner portals, employee apps, legacy services and third-party integrations. Make a list of every login page your company exposes to the internet. Retire the ones nobody uses.
Use the free detection pack, with care
Researcher MalwareMustDie released an open-source ARTEX detection-signature pack on the same Friday as the FBI advisory, according to BleepingComputer. It covers:
- Suspicious network requests
- Identity abuse
- Cloud activity
- Malicious files
- Automated attack patterns
A companion playbook sorts indicators by confidence level. It warns that IP addresses and other infrastructure indicators can go stale or trigger false positives. Use the rules as one input, not a final answer.
Plan for speed
If some intrusions reach privileged access in under three hours, a morning-after review is too slow. Set alerts for new admin accounts, changed credentials and big data transfers. Decide in advance who gets called and who can lock accounts.
Report suspicious activity
Organizations that see ARTEX-related activity are encouraged to report it to the FBI or its Internet Crime Complaint Center (IC3).
Budget honestly
Identity monitoring, logging and permission reviews cost time and money. The FBI's description of ARTEX suggests these costs are becoming a basic expense, not an extra. Start with the cheapest steps: permission clean-up, retiring unused apps and turning on alerts.
Source: BleepingComputer report on the FBI ARTEX advisory; Microsoft Security Blog, October 21
Frequently Asked Questions
Q: What is ARTEX AI?
A: ARTEX is an open-source, AI-powered penetration-testing framework written in Go. The FBI says threat actors have used it to automate reconnaissance, exploitation, privilege escalation and data theft. It connects to outside AI models rather than using its own.
Q: What did the FBI ARTEX advisory say?
A: The FBI's FLASH advisory said multiple threat actors have used ARTEX since at least August 2026. It said some intrusions went from initial targeting to privileged access in under three hours. The advisory is 13 pages long and includes indicators of compromise.
Q: Who is Storm-0940, and how is Storm-0940 ARTEX activity linked to ConnectWise?
A: Microsoft describes Storm-0940 as a China-state-aligned threat actor. Microsoft said it used ARTEX to speed up attacks tied to the ConnectWise Cloud Backup compromise, and had used the tool since at least September.
Q: What is Vapor Panda, and what did the Vapor Panda ARTEX campaign do?
A: CrowdStrike tracks Vapor Panda as a Chinese cyberespionage group. It said the group used ARTEX in a campaign that began October 17 and had targeted more than 200 organizations in at least 14 countries by November 4. At least nine were compromised.
Q: Did the Vapor Panda campaign breach 200 organizations?
A: No. CrowdStrike said more than 200 were targeted and at least nine were compromised. Targeting is not the same as a successful breach.
Q: Which AI models does ARTEX work with?
A: According to the FBI, ARTEX integrates with Anthropic Claude, OpenAI GPT, DeepSeek, GLM, Grok and Kimi. The framework itself has no built-in model.
Q: Is ARTEX still available after it was removed from GitHub?
A: Yes. The developer, Autumn-27, removed the repository on October 8. The FBI said ARTEX remains available through forks, archived copies, localized versions and a commercial edition called ARTEX PRO.
Q: How many ARTEX consoles are exposed online?
A: Researchers found more than 1,600 publicly exposed ARTEX consoles as of early November. The actual number of deployments was believed to be significantly higher.
Q: How can a small business protect itself from ARTEX-style attacks?
A: Follow the FBI's guidance. Secure internet-facing apps, restrict cloud admin access and review app permissions and authentication logs. Also add rate limits to sensitive APIs and watch for odd credential changes and large data transfers.
Q: Where do I report suspected ARTEX activity?
A: The FBI encourages organizations that see suspicious activity tied to ARTEX to report it to the FBI or its Internet Crime Complaint Center (IC3). Check the FBI advisory's indicators first, but remember some IP addresses may have changed.
Additional Links
Flowgrammer.ca – 💡 Unlock AI-Powered Automation! Let flowgrammer.ca handle complex workflows so you can focus on growing your business.
Flowgrammers.ca – 💡 Join Toronto's Fastest Growing AI and Business Automation Community.
- SRED Funding Solutions – Maximize your R&D tax credits and secure funding for your innovation.
- Silicon Scoop Podcast – Stay Informed, Stay Ahead!
- Startup Talk Podcast
- Flowgrammer Youtube
- Startup Investor Drinks – Join Toronto’s Startup Community and connect with fellow startup enthusiasts and investors.
- TorontoStarts Startup and Entrepreneur Community – Become a part of Toronto’s thriving startup ecosystem.
- Workhaus – Discover flexible workspace solutions for your business needs.
- Work with The Startup Coach – Get expert guidance and mentorship for your startup journey.
- Reach out to Jon Irwin on LinkedIn – Connect with our co-host and technology expert.
