Glossary term
Cyber espionage
Cyber espionage is the use of computer networks to secretly gather information that someone wants to keep private.
Quick links
Press Enter to search, Esc to close
Significant cyber incident
In short
On 8 January 2025 Japan's National Police Agency (NPA) and the National center of Incident readiness and Strategy for Cybersecurity (NISC) said that cyberattack campaigns against Japanese organizations since about 2019 were run by an actor called MirrorFace. They described three campaigns, by email and through flaws in internet-facing devices, aimed at information on national security and advanced technology. The agencies assessed the activity as organized, with suspected Chinese involvement.
On 8 January 2025 the NPA and NISC published an alert about MirrorFace, also called Earth Kasha. They said cyberattack campaigns against organizations, businesses and individuals in Japan, from about 2019 to the present, were conducted by this actor.
The alert splits the activity into three campaigns. Campaign A (2019 to 2023) used emails with malicious attachments against think tanks, government including retirees, politicians and media. Campaign B (from around 2023) used flaws in internet-connected devices to enter networks, mainly at semiconductor, manufacturing, telecommunications, academic and aerospace targets. Campaign C (from around June 2024) used emails with links to download malware, aimed at academia, think tanks, politicians and media.
The NPA and NISC assessed the campaigns as a series of organized cyberattack activities with suspected Chinese involvement, mainly aimed at stealing information on national security and advanced technology. The alert does not name the individual victims.
The British broadcaster LBC, reporting the NPA statement, said Japanese authorities linked more than 200 attacks over five years to MirrorFace and that targets included the foreign and defense ministries and the space agency. The NPA text opened for this page does not give those numbers or name those bodies. LBC also calls the attacks “Chinese government-led”, while the NPA says “suspected Chinese involvement”, so sources differ on how firm the attribution is.
The NPA and NISC date the start of the campaigns attributed to MirrorFace to about 2019.
Campaign A: emails with malicious attachments sent to think tanks, government, politicians and media. Some compromise from it continued until about May 2024.
The NPA says attackers were already using Windows Sandbox to run malware out of sight of security tools.
Campaign B: VPN device flaws or credentials obtained by some means are used to enter networks of semiconductor, manufacturing, telecom, academic and aerospace targets.
The NPA still finds compromise caused by Campaign A infections.
Campaign C starts: emails with links to download malware. Attackers also begin using the VS Code development tunnel feature (at the latest).
The NPA and NISC publish the alert with methods, traces and detection advice. LBC reports it the same day.
The NCO website lists the English alert "Cyberattacks by MirrorFace".
The NPA's report on threats in 2025 repeats the assessment of organized activity likely involving China.
The NPA alert and its two technical annexes describe three ways in. After entry the attackers used a few tricks that make their traces hard to find.
Emails used Gmail or Outlook addresses, or genuine third-party addresses that had been abused. Subjects followed the security news of the day, such as the Japan-US alliance, the Taiwan Strait and the war in Ukraine. Opening the file led to infection with a tool called LODEINFO, and in some cases to LilimRAT or NOOPDOOR (Phishing, T1566).
The NPA says attackers entered through VPN devices, either through flaws or through credentials they had obtained. The flaws it names are in Array Networks (CVE-2023-28461), Fortinet FortiOS and FortiProxy (CVE-2023-27997) and Citrix ADC and Gateway (CVE-2023-3519) (Exploit Public-Facing Application, T1190). Tunnelling tools and web shells were placed on the devices (Web Shell, T1505.003). Attackers then reached Microsoft 365 and virtualisation servers, and took virtual machine images.
Emails carried a link to download malware, with subjects such as interview requests and international affairs. The NPA says the bodies of the emails look like altered copies of earlier genuine emails, so they feel normal. Infection was with a tool called ANEL.
Windows Sandbox is a built-in temporary Windows environment. The NPA says attackers set it up to run malware inside it, which lets the malware avoid antivirus and EDR tools. Traces inside it are erased when the PC is shut down or restarted. From June 2024 at the latest attackers also used the development tunnel feature of Visual Studio Code to run commands remotely.
Who attributed it, and how firm it is
Attributed by the NPA and NISC, 8 January 2025, to MirrorFace (also called Earth Kasha). They assess the campaigns as organized cyberattack activity with suspected Chinese involvement. LBC's report of the same statement calls the attacks "Chinese government-led", which is stronger wording than the NPA text, so sources differ in tone. The NPA text does not name a Chinese government body.
The NPA and NISC say the campaigns aimed at information on national security and advanced technology in Japan, and name target groups: think tanks, government including retirees, politicians, media, academia, and semiconductor, manufacturing, telecommunications and aerospace firms. The NPA text opened does not give a count of victims or say what was taken. LBC reports figures of more than 200 attacks and names the foreign and defense ministries and the space agency; those figures are not in the NPA pages opened, so treat them as unconfirmed here.
The NPA says Campaign C email text looks like altered copies of genuine earlier emails. Staff cannot rely on "it reads normally" as a check.
Campaign B entered through VPN devices. Patch speed and credential hygiene on these devices matter as much as on servers.
The NPA advises replacing a device that was ever compromised, since a patch may not remove what was left behind.
Windows Sandbox and VS Code tunnels are normal Microsoft features. Disable or watch them where the business does not need them.
The NPA stresses central, wide and long log storage because attackers can erase logs on the systems they enter.
The campaigns run from about 2019. Compromise from Campaign A was still found in May 2024.
List your VPN and gateway products and check whether CVE-2023-28461, CVE-2023-27997 and CVE-2023-3519 were ever open on them. Replace any device that was compromised.
Look for logins from abroad when staff are all in one country, odd user agents, and entries with loopback addresses as the source.
If the business does not use Windows Sandbox, disable it and alert if it turns on again.
Check whether VS Code is installed without a reason and watch for traffic to development tunnel services.
Treat password-protected zip files, VHD or ISO attachments and download links in unexpected mail as suspect, as the NPA advises.
Do not enable content in Office files from unexpected senders. Confirm with the sender by another route.
Send logs from VPNs, directory servers and Microsoft 365 to one store that attackers cannot edit.
Know how to report to the NPA, NISC (now NCO) or JPCERT/CC if you see these signs.
The NPA and NISC use the name MirrorFace, also Earth Kasha, for the actor behind the campaigns. They assess its activity as organized, with suspected Chinese involvement.
The NPA and NISC say the aim was mainly information related to national security and advanced technology in Japan.
By email with attachments (Campaign A), by flaws in internet-facing devices such as VPNs (Campaign B), and by email with download links (Campaign C).
The NPA says it used Windows Sandbox to hide malware from security tools and, from June 2024, the VS Code development tunnel feature for remote commands.
It comes from the LBC report of the NPA statement. The NPA pages opened for this page do not give it, so it is not confirmed here.
Take control of your digital security with an exclusive demo of our AI-driven threat intelligence platform.