Security

Chinese AI tool linked to hacks in Japan – tracking attribution and response

The Chinese AI tool Artex is linked to breaches at Japanese firms and more than seven South Korean financial institutions. A statement from the National Cybersecurity Office regarding new mandatory reporting requirements for AI-driven breaches would confirm the first operational phase of the new command structure.

Chinese AI tool linked to hacks in Japan – tracking attribution and response

The story so far

edited Oct 24 · 2 paragraphs changed

What is happening

Japanese infrastructure and financial firms face increased risk from automated cyberattacks as security researchers link a Chinese-made artificial intelligence tool to domestic data breaches 910. The tool, known as Artex or Autumn-27, is an autonomous agent that connects large language models to network testing software to automate the discovery of vulnerabilities 12. Cybersecurity firm GMO Cybersecurity reported that the tool has been used to target Japanese companies 910. This follows a warning from Japanese authorities on October 8, 2026, that the country is experiencing an unusually high volume of corporate data breaches this year due to AI lowering barriers for attackers 4.

The developer of Artex converted the project to closed-source and stopped public updates on October 8, 2026 12. The GitHub repository hosting the code was removed following reports that the tool was used in breaches at more than seven financial institutions in South Korea 123. The South Korean Financial Services Commission stated these attacks affected more than 68,000 people 3. While the developer claimed the tool was intended for legitimate security risk testing, the project was pulled after the reports of misuse in the banking sector 12.

Prime Minister Sanae Takaichi designated the National Cybersecurity Office as a central command center on October 23, 2026, to strengthen defenses against sophisticated cyberthreats 21. The move follows a proposal from the ruling Liberal Democratic Party to establish a dedicated headquarters to coordinate the national response to recent large-scale data breaches 21. Takaichi stated the government will develop a new framework to accelerate defensive measures across all relevant ministries and agencies 21.

added Oct 24

Prime Minister Takaichi designated a new cybersecurity command center on October 23, 2026 [21].

Microsoft attributed a global cyber campaign involving the Artex framework to the China-aligned threat actor Storm-0940 on October 21, 2026 1819. The actor used the tool to automate scanning and reconnaissance against hundreds of organizations in sectors including aerospace, defense, and semiconductors 1718. Between October 10 and 14, 2026, Storm-0940 infrastructure conducted an automated password-spray campaign targeting 327 accounts, peaking at 2,900 failed sign-in attempts per minute on October 11 18. In several instances, the time from initial targeting to obtaining privileged cloud access was less than three hours 1218.

Storm-0940 used stolen credentials to authenticate to customer Microsoft 365 environments by targeting identities of multiple ConnectWise cloud applications on October 15, 2026 1820. The actor targeted more than 2,250 tenants, with at least 162 confirmed as compromised 1720. ConnectWise shut down its Cloud Backup service on October 17, 2026, after identifying the unauthorized access 1819.

The developer of Artex converted the project to closed-source and stopped public updates on October 8, 2026 12. The GitHub repository hosting the code was removed following reports that the tool was used in breaches at more than seven financial institutions in South Korea 123. The South Korean Financial Services Commission stated these attacks affected more than 68,000 people 3. While the developer claimed the tool was intended for legitimate security risk testing, the project was pulled after the reports of misuse in the banking sector 12.

What to watch

A public attribution of the attacks to the Artex tool by Japan’s National Center of Incident Readiness and Strategy (NISC) would clarify the scope of the threat to critical infrastructure. The Japanese government may announce sanctions or export controls targeting the tool or its developers in response to the breaches. A statement from Chinese authorities confirming the removal or restriction of the tool from domestic platforms would indicate a regulatory response to the misuse of the software. The publication of a final report confirming mitigation measures and the closure of the investigation would signal the end of the current response phase.

A formal announcement detailing the specific authorities and budget of the new National Cybersecurity Office command center would clarify the extent of its oversight over private infrastructure providers. A cabinet decision to adopt the Liberal Democratic Party's proposal for a dedicated headquarters would settle whether the response remains within existing offices or moves to a new agency. A statement from the National Cybersecurity Office regarding new mandatory reporting requirements for AI-driven breaches would confirm the first operational phase of the new command structure.

added Oct 24

The new command center development necessitates tracking its specific authorities and the LDP proposal status [21].

Read the full brief · how we got here

How we got here

Japan adopted a basic policy on July 31, 2026, to improve early threat detection for critical infrastructure following a cyberattack on the frozen food manufacturer Nichirei 5. The policy shift occurred as the government sought to address vulnerabilities in essential services 5. Cybersecurity firms including Crowdstrike and government agencies in South Korea first identified the link between the Artex tool and financial sector breaches in early October 2026 23. The tool was designed to use models like ChatGPT and Claude to automate penetration testing, but its autonomous capabilities were redirected for data theft 111.

References · 14
  1. 1en.sedaily.com — China-Made AI Hacking Tool ARTEX Goes Private After Korean Bank Attacks
  2. 2manilatimes.net — China's Artex AI agent made closed - The Manila Times
  3. 3progressiverobot.com — ARTEX AI Pulled: Surprising Misuse Warning After Korea Hacks
  4. 4japantimes.co.jp — Japan and cybersecurity: What's behind recent attacks and what's being ...
  5. 5japantimes.co.jp — Japan adopts cybersecurity policy for early threat detection
  6. 9letsdatascience.com — GMO Cybersecurity Links China-Built AI Tool ARTEX to Japan Breaches
  7. 10cryptobriefing.com — GMO Cybersecurity links Chinese AI tool ARTEX to breaches at Japanese firms
  8. 11thehackernews.com — ARTEX AI Pentesting Tool Used in Data Theft Attacks on South Korean ...
  9. 12malware.news — AI-driven, large-scale nation-state campaign using ARTEX enables supply chain compromise - Malware News - Malware Analysis, News and Indicators
  10. 17thehackernews.com — China-Linked Hackers Use ARTEX AI Agents in ConnectWise Supply Chain Attack
  11. 18microsoft.com — AI-driven, large-scale nation-state campaign using ARTEX enables supply chain compromise | Microsoft Security Blog
  12. 19therecord.media — Microsoft blames ConnectWise attacks on Chinese state hacker using ARTEX AI agents | The Record from Recorded Future News
  13. 20cybersecuritynews.com — China-Linked Hackers Use ARTEX AI to Target 2,250 Microsoft 365 Tenants After ConnectWise Breach
  14. 21nippon.com — Japan to Boost Cybersecurity Measures against Data Breaches | Nippon.com

What would close this

0 of 4 met
Japan’s National Center of Incident Readiness and Strategy (NISC) publicly attributes the attacks to the Artex AI tool.open
The Japanese government announces sanctions or export controls targeting the tool or its developers.open
Chinese authorities issue a statement confirming removal or restriction of the tool from domestic platforms.open
A final report confirming mitigation measures and closure of the investigation is published.open
If none of these happen, this page closes automatically on .

Timeline

4newest first
Link

Japan to establish cybersecurity command center following Artex AI breaches

Prime Minister Sanae Takaichi announced on October 23, 2026, that Japan will accelerate the development of a new cybersecurity framework to counter increasingly sophisticated AI-driven threats. The move follows a formal proposal from the ruling Liberal Democratic Party to establish a dedicated headquarters to coordinate the national response to recent large-scale data breaches.

  • Prime Minister Takaichi designated the National Cybersecurity Office as the central command center to strengthen defenses against sophisticated cyberthreats 1.
  • The government response follows an October 9, 2026, assessment by ruling party officials that the country is in a "state of emergency in cyberspace" after breaches at more than 20 major companies 356.
  • Recent attacks linked to the Artex AI tool have compromised infrastructure at cloud provider IDC Frontier, affecting 495 clients including JR East and Times Car 35.
Full analysiswritten from the full pages behind this update

What happened

Prime Minister Sanae Takaichi designated the National Cybersecurity Office as the central command center to strengthen national defenses on October 23, 2026 1. This administrative shift follows a formal proposal from Masaaki Taira, head of the Liberal Democratic Party’s (LDP) Headquarters for National Cybersecurity Strategy, who characterized the current volume of data breaches as a matter of urgent public concern 1. Takaichi stated the government will now develop a new framework to accelerate defensive measures across all relevant ministries and agencies 1. The move responds to a series of large-scale personal data breaches that Takaichi described as increasing in both scale and sophistication 1. While the Prime Minister’s office coordinates this new command structure, the LDP proposal specifically called for a dedicated headquarters to manage the response to these cyberattacks 1.

Why it matters

The centralization of cybersecurity under the Prime Minister’s office establishes a new legal and operational hierarchy for Japanese digital defense 1. By elevating the National Cybersecurity Office to a command center, the government aims to unify the response of disparate ministries that have previously managed infrastructure security in silos 1. For the private sector, this shift signals a transition from voluntary compliance to a state-led defensive posture, as Takaichi noted that the scale of recent threats has reached a level requiring direct leadership from the prime minister’s office 1. The political urgency cited by Taira suggests that the government is prioritizing the restoration of public confidence following the breach of personal data across multiple sectors 1. The effectiveness of this command center will be measured by its ability to integrate the "necessary framework" Takaichi promised to develop, which will dictate how future AI-driven incidents are reported and mitigated 1.

What's next

The government is scheduled to begin the development of the new cybersecurity framework immediately under the leadership of the Prime Minister's office 1. A formal announcement detailing the specific authorities and budget of the new National Cybersecurity Office command center would clarify the extent of its oversight over private infrastructure providers. The LDP's proposal for a dedicated headquarters remains under review; a cabinet decision to adopt this specific structure would settle whether the response remains within existing offices or moves to a new agency 1. Future diplomatic coordination may be informed by the Trilateral Cyber Dialogue framework, which previously established channels between Japan, China, and South Korea for discussing cyber strategies and capacity building 2. A statement from the National Cybersecurity Office regarding new mandatory reporting requirements for AI-driven breaches would confirm the first operational phase of the new command structure.

Link

China-linked Storm-0940 uses Artex AI to breach hundreds of organizations in under three hours

A China-state aligned actor, Storm-0940, has been using the Artex agentic AI framework since September 2026 to automate global cyberattacks, achieving privileged cloud access in less than three hours 1. The tool's commercial availability and its deployment by state actors mark a shift toward high-speed, automated supply chain compromises 1.

  • Storm-0940 deployment: The actor used Artex to automate penetration testing against hundreds of organizations across multiple industries globally 1.
  • Rapid exploitation: In several instances, the time from initial targeting to obtaining privileged access to cloud infrastructure was recorded at under three hours 1.
  • Supply chain risk: The campaign specifically enables large-scale supply chain compromises through automated vulnerability discovery 1.
Full analysiswritten from the full pages behind this update

What happened

Microsoft attributed a global cyber campaign involving the agentic AI framework ARTEX to the China-aligned threat actor Storm-0940 on October 21, 2026 57. The actor used the Go-based framework to automate scanning, reconnaissance, and password spraying against hundreds of organizations across sectors including aerospace, defense, and semiconductors 67. Between October 10 and 14, 2026, Storm-0940 infrastructure conducted an automated cross-tenant password-spray campaign targeting 327 accounts, peaking at 2,900 failed sign-in attempts per minute on October 11 6.

On October 15, 2026, the actor gained access to the identities of multiple ConnectWise cloud applications 6. Storm-0940 used these stolen credentials to authenticate to customer Microsoft 365 environments without requiring interactive user sign-ins, bypassing multi-factor authentication by acting as the trusted service principal 56. ConnectWise shut down its Cloud Backup service on October 17, 2026, after identifying unauthorized access to application identities 5. Microsoft reported that in several instances, the time from initial targeting to obtaining privileged cloud access was less than three hours 7.

The ARTEX framework, developed by a Chinese cybersecurity engineer, was converted to closed-source on October 8, 2026, following reports of its abuse in attacks against South Korean financial institutions 8. The developer stated the tool was originally intended for authorized security risk testing and research 8. Zscaler researchers documented that the framework has been used to coordinate multiple AI models, including DeepSeek and GLM, to automate tasks from reconnaissance to data exfiltration 9.

Why it matters

The use of agentic AI allows threat actors to operate at "machine speed," attempting multiple breach paths simultaneously and working without the constraints of human schedules 78. This automation significantly reduces the window for defenders to detect and respond to intrusions 6. In the ConnectWise incident, attackers targeted more than 2,250 tenants, with at least 162 to 173 tenants confirmed as compromised 59.

The campaign highlights a shift toward automated supply chain compromises through the abuse of OAuth applications and service principals 17. Because the attackers authenticated as the application itself using stolen secrets, they inherited permissions previously granted by customers, such as the ability to read mail and files for backup purposes 5. Huntress observed the actor using these identities to assign directory roles, including Exchange Administrator and Global Administrator, to establish persistent access 5. While Microsoft identified follow-on activity including reconnaissance and credential changes, the company stated that successful authentication does not necessarily indicate that customer data was exfiltrated 79.

What's next

ConnectWise has disabled seven application identities covering Cloud Backup and integrations for Teams, Exchange Online, and Microsoft Graph 5. The company recommends that affected connections remain disabled until formal restoration guidance is issued 9.

Security teams are advised to review service principal sign-ins and investigate unexpected changes to application credentials or privileges 7. Microsoft recommends preserving evidence before deleting applications, as attackers may have added credentials to other authorized integrations to maintain access after the primary connection is severed 9. A statement from the Japanese National Center of Incident Readiness and Strategy (NISC) or the Ministry of Internal Affairs and Communications would clarify if domestic critical infrastructure was among the hundreds of organizations targeted by Storm-0940 during this campaign.

MinorLink

Chinese security firm ThreatBook commercializes Artex AI tool following Japanese and South Korean breaches

The Artex AI tool, recently linked to breaches at Japanese infrastructure and South Korean financial firms, is being integrated into a new commercial cloud platform called RedAgents.ai by Chinese cybersecurity firm ThreatBook 34. This transition to a paid service follows the developer's move to take the original code private after reports of its misuse in automated attacks 24.

  • Commercialization of Artex: ThreatBook CEO Xue Feng announced on October 16, 2026, that a commercial version called ARTEX PRO will lead the new RedAgents.ai platform 34.
  • Developer Ties: The original creator of Artex is a threat detection engineer at ThreatBook who is now leading the commercial project 4.
  • Japanese Target Scope: Forensic analysis by GMO Cybersecurity released on October 19, 2026, confirmed Artex was used to target approximately 20 Japanese companies across sectors including retail, education, insurance, and logistics 1.
  • South Korean Investigation: The National Police Agency's Cyber Investigation Bureau is investigating the tool's role in the theft of 100,000 records from 16 South Korean financial institutions between late September and early October 2026 3.
Link

When we started following: Artex AI developer takes tool private following cyberattacks on South Korean financial institutions

The developer of the Chinese-made AI penetration-testing tool Artex, known as Autumn-27, converted the project to closed-source and halted public updates on October 8, 2026 12. The move followed reports from cybersecurity firm Crowdstrike and the South Korean government linking the tool to breaches at more than seven financial institutions 23. South Korea’s Financial Services Commission reported that the attacks affected more than 68,000 people 3.

Artex is an autonomous agent designed to link large language models like ChatGPT and Claude to automate network vulnerability testing 12. While the developer stated the tool was intended for security risk testing and opposed its illegal use, the GitHub repository hosting the code was taken down after reports of its misuse in the banking sector 12. In Japan, Prime Minister Sanae Takaichi adopted a basic policy on July 31, 2026, to enhance early threat detection for critical infrastructure following a cyberattack on frozen food maker Nichirei 5. Japanese authorities warned on October 8, 2026, that the country faces an unusually high volume of corporate data breaches this year as AI lowers barriers for attackers 4.