Google Threat Intelligence Group (GTIG) has uncovered a long-running cyberespionage campaign, active since at least November 2022, conducted by the China-linked threat actor APT24 (also known as Pitty Tiger). The campaign's primary objective is intelligence gathering, with a strong focus on targets in Taiwan. A key component of this operation is a previously undocumented custom malware downloader named BadAudio. This malware is a highly obfuscated first-stage payload designed to establish a foothold on victim systems before delivering more powerful tools like Cobalt Strike. APT24 has demonstrated significant operational evolution, shifting from broad watering hole attacks to more targeted supply chain compromises and spear-phishing, indicating a persistent and sophisticated adversary focused on long-term intelligence collection.
APT24 is a state-sponsored threat group with a history of targeting organizations for intellectual property theft, active since at least 2008. This latest campaign showcases their continued investment in custom tooling and adaptive tactics. The operation has progressed through several phases:
BadAudio is a first-stage downloader written in C++ and serves as the initial entry point into a victim's network. Its primary features are geared towards stealth and evasion:
T1574.001 - Hijack Execution Flow: DLL Search Order Hijacking, where a legitimate application is tricked into loading the malicious DLL instead of a legitimate one.T1199 - Trusted Relationship: Exploited by compromising a digital marketing firm to attack its customers (supply chain attack).T1189 - Drive-by Compromise: Used in the initial phase by compromising legitimate websites (watering hole).T1566.002 - Phishing: Spearphishing Link: Employed to deliver malicious archives via email.T1574.001 - Hijack Execution Flow: DLL Search Order Hijacking: The primary execution vector for the BadAudio DLL.T1027 - Obfuscated Files or Information: Utilized control flow flattening to hinder analysis.T1071.001 - Application Layer Protocol: Web Protocols: Used HTTP for C2 communications.T1573.001 - Encrypted Channel: Symmetric Cryptography: Used AES to encrypt C2 traffic.T1059.003 - Command and Scripting Interpreter: Windows Command Shell: Likely used by the second-stage payload (Cobalt Strike) for execution.The campaign's primary goal is cyberespionage. By targeting sectors like government, healthcare, telecommunications, and construction, APT24 seeks to steal valuable intellectual property, government secrets, and other sensitive data that align with the strategic interests of the Chinese state. The compromise of a digital marketing firm represents a significant supply chain risk, exposing a large number of downstream organizations to infection. For targeted organizations in Taiwan, this long-term, persistent intrusion could lead to a substantial loss of sensitive national and commercial information.
No specific file hashes or C2 domains were provided in the source articles.
| Type | Value | Description | Context | Confidence |
|---|---|---|---|---|
| file_name | *.dll |
Monitor for newly dropped DLLs in directories where legitimate applications load their libraries. | File Integrity Monitoring, EDR | medium |
| command_line_pattern | rundll32.exe <suspicious.dll>,<export_name> |
A common method for executing malicious DLLs. Look for executions of unsigned or untrusted DLLs. | EDR, Windows Event ID 4688, Sysmon Event ID 1 | medium |
| network_traffic_pattern | HTTP requests with encrypted data in cookies | Unusual or large amounts of data being sent in HTTP cookie headers could indicate BadAudio C2 communication. | Network Security Monitoring, IDS/IPS | medium |
| process_name | [legitimate_process.exe] |
Monitor legitimate processes for suspicious network connections or for loading unsigned DLLs from non-standard paths. | EDR, Process Monitoring | high |
Detecting BadAudio requires a defense-in-depth strategy that combines network and endpoint monitoring.
Endpoint Detection (EDR): An EDR solution is critical for detecting techniques like DLL search-order hijacking. Configure EDR to alert on legitimate processes loading unsigned or newly created DLLs. Memory scanning can also detect the in-memory execution of payloads like Cobalt Strike. This is an application of D3-PA: Process Analysis.
Network Monitoring: Analyze outbound HTTP traffic for anomalies. Since BadAudio uses cookies for C2, look for POST requests with unusually large or encrypted cookie values. Decrypting SSL/TLS traffic is essential for this analysis. This aligns with D3-NTA: Network Traffic Analysis.
Threat Hunting: Proactively hunt for signs of DLL hijacking. Query systems for applications that have DLLs with the same name in both the application directory and a system directory (e.g., System32), as this is a common condition for hijacking.
Defending against a sophisticated actor like APT24 requires robust security hygiene and advanced controls.
Use application control to prevent the loading of unauthorized or unsigned DLLs, which would block the DLL search order hijacking technique.
Use web filtering and SSL/TLS inspection to block connections to known malicious domains and detect anomalous C2 traffic.
Educate users to recognize and report suspicious pop-ups and phishing attempts.
To directly counter the core execution method of the BadAudio malware, organizations should implement robust executable allowlisting, with a specific focus on DLLs. Using technologies like Windows Defender Application Control (WDAC), security teams can create policies that only permit known, signed, and trusted DLLs to be loaded by applications. This fundamentally breaks the DLL search order hijacking technique (T1574.001) used by APT24, as the operating system would prevent the legitimate application from loading the malicious BadAudio DLL. The policy should be configured in enforcement mode, starting with critical systems and servers, and then gradually rolled out to workstations. While this requires a significant initial investment in baselining and policy creation, it is one of the most effective controls against this class of malware.
For detection of BadAudio and its second-stage payloads like Cobalt Strike, continuous process analysis on endpoints is essential. Security teams must leverage an EDR solution to monitor for anomalous process behaviors. Specifically for this threat, configure detection rules for: 1) Legitimate, signed processes (e.g., explorer.exe) loading unsigned DLLs from non-standard directories. 2) A process establishing a network connection with high-entropy data in the HTTP cookie header, which is characteristic of BadAudio's C2. 3) In-memory injection of threads into other processes, a hallmark of Cobalt Strike's beacon. By establishing a baseline of normal process activity and alerting on these specific deviations, security teams can detect the infection chain even if the initial malware files are heavily obfuscated.

Cybersecurity professional with over 10 years of specialized experience in security operations, threat intelligence, incident response, and security automation. Expertise spans SOAR/XSOAR orchestration, threat intelligence platforms, SIEM/UEBA analytics, and building cyber fusion centers. Background includes technical enablement, solution architecture for enterprise and government clients, and implementing security automation workflows across IR, TIP, and SOC use cases.
Help others stay informed about cybersecurity threats