Kaspersky uncovers ShadowCloud espionage campaign on government networks
Kaspersky's Global Research and Analysis Team (GReAT) has disclosed Operation ShadowCloud, an active cyber-espionage campaign directed at government organisations across Southeast Asia, Africa and Central Asia. The campaign uses a combination of legitimate remote-management software and cloud services to deliver malware and exfiltrate data while masking activity within ordinary business traffic.
Researchers found circumstantial links to Chinese-speaking threat groups, specifically FamousSparrow and CopperTurtle, though Kaspersky stopped short of a definitive attribution. The campaign targets Windows environments and appears to leverage existing footholds to gain initial access, after which attackers deploy MeshCentral, a legitimate web-based remote-management tool, to stage the follow-on infection.
How the attack chain works
Once MeshCentral is running on a compromised host, it delivers three files: a legitimate signed executable, a malicious DLL, and an encrypted malware payload. The attackers exploit DLL sideloading, a technique in which a trusted program is tricked into loading a malicious library, to decrypt and execute the payload. That payload launches one of two previously undocumented backdoors, dubbed GitDoor and GraphDoor.
Both backdoors use cloud services including GitHub, Microsoft SharePoint and the developer messaging platform Ably as command-and-control channels. The choice of widely trusted platforms makes the traffic harder to distinguish from routine API calls. Once active, the backdoors grant operators broad capability: keystroke logging, screenshot capture, file search and exfiltration, credential extraction from memory, manipulation of Windows services and firewall rules, and even retrieval of WhatsApp database decryption keys.
Saurabh Sharma, lead security researcher at Kaspersky GReAT, said: "This attack combines several tactics that may appear familiar but may be hard to detect using traditional network and file-based scanning. Organisations should always monitor the use of remote-management tools, cloud APIs, unusual DLL loading, and memory-based execution rather than relying only on traditional malware and network controls."
Market and threat landscape context
The abuse of legitimate cloud infrastructure for command and control has become a defining characteristic of state-aligned and state-adjacent espionage operations over the past several years. By routing communications through GitHub or SharePoint tenants, attackers benefit from the same TLS certificates and allowlisting decisions that security teams apply to routine SaaS traffic, substantially raising the detection bar for perimeter and network-inspection tools.
The government sector in Southeast Asia and Central Asia has been a persistent focus for espionage campaigns, with several advanced persistent threat groups attributed to Chinese-speaking actors by Western governments and private threat-intelligence firms. The UK National Cyber Security Centre, alongside counterparts in the Five Eyes alliance, has issued guidance on detecting living-off-the-land and cloud-abuse techniques, recommending behavioural endpoint detection and centralised logging as primary countermeasures.
Kaspersky itself operates in a constrained commercial environment in several Western jurisdictions following US government restrictions imposed in 2024. The firm continues to publish threat-intelligence research through its Securelist platform, which carries independent analytical value regardless of the commercial situation. Security teams outside the US and UK government sectors that subscribe to Kaspersky threat feeds will find the indicators of compromise from this investigation directly actionable.
The full technical report, including indicators of compromise and YARA rules for GitDoor and GraphDoor, is available on Securelist. The near-term priority for affected governments and their technology suppliers will be auditing remote-management tool deployments and reviewing cloud-service API egress logs for anomalous patterns consistent with the described behaviour.