Home › Intelligence › Brief
BREACH BRIEF 🟠 High ThreatIntel

HTTP/3‑Powered RATs Use QUIC to Evade Detection in Southeast Asian Government Networks

Unit 42 reports a cyber‑espionage campaign that has compromised Southeast Asian government systems using custom RATs that communicate via HTTP/3 over QUIC, bypassing traditional TCP‑focused defenses. The actors exploited unpatched Exchange servers and phishing to harvest credentials and exfiltrate data, demonstrating a gap in network‑traffic monitoring controls.

Verisq™ Intelligence · 📅 November 17, 2026 · 📰 unit42.paloaltonetworks.com
Get the weekly threat digestBreaches, exploited vulnerabilities and advisories that matter to your vendors, every Friday. Free.
🟠
Severity
High
TI
Type
ThreatIntel
🎯
Confidence
High
🏢
Affected
1 sector(s)
✅
Actions
2 recommended
📰
Source
unit42.paloaltonetworks.com

HTTP/3‑Powered RATs Use QUIC to Evade Detection in Southeast Asian Government Networks

What Happened — Unit 42 uncovered a multi‑stage cyber‑espionage campaign that has been targeting government entities across Southeast Asia since mid‑2025. The actors deployed two previously undocumented remote‑access Trojans—PerfectPioneer RAT and its successor KUKURAT—both of which use HTTP/3 over QUIC for command‑and‑control, a protocol that bypasses many TCP‑centric inspection tools. Initial footholds were gained through exploitation of unpatched Microsoft Exchange/web servers and spear‑phishing emails with decoy PDFs, after which the RATs performed credential harvesting, data exfiltration and extensive stealth techniques such as synthetic noise generation and a safety‑switch that blocks noisy commands.

Unit 42’s threat research team disclosed a multi‑stage cyber‑espionage operation that has been probing Southeast Asian government agencies since at least July 2025. The campaign is catalogued as TGR‑STA‑0049—also known by the aliases REF7707, Earth Alux and Ink Dragon—and is distinguished by the use of highly customized payloads engineered for stealthy intelligence gathering while evading conventional detection tools. According to lead analyst Tom Fakterman, “the group’s hallmark is targeting government entities with advanced, low‑noise backdoors.”

The actors behind TGR‑STA‑0049 previously fielded the Squidoor backdoor (also called FINALDRAFT), a Windows and Linux implant active since 2023. In the current wave they introduced two previously undocumented remote‑access trojans: PerfectPioneer RAT, first observed in July 2025, and its modular successor KUKURAT, which appeared in the wild in March 2026. Over a nine‑month period the researchers tracked iterative refinements across six PerfectPioneer variants—newhttp3, PerfectRat, Pioneer, RatAAkk‑tt, rat_http3_ls and rat_http3_cc—each incorporating new features while preserving a common code base.

Three capabilities set these families apart. First, both RATs employ covert HTTP/3 (QUIC) communications over UDP, a protocol that can slip past deep‑packet inspection tuned for TCP traffic. Second, PerfectPioneer includes an “isDangerousCommand” validation engine that blocks operators from issuing noisy or destructive commands that might trigger alerts. Third, KUKURAT adds a custom spoofing engine that fabricates benign‑looking network, registry and file activity, flooding the endpoint’s behavioral baseline with synthetic noise to hide indicators of compromise.

Infection is achieved through a dual‑track approach. The primary vector is server exploitation, where attackers compromise unpatched Microsoft Exchange and other web servers to drop web shells such as ASPXSpy, Godzilla, Bysunue.aspx and Suo5. These shells are then used to launch the PerfectPioneer payload and conduct initial reconnaissance, as illustrated in the team’s Figure 1. A secondary, less frequent vector involves phishing campaigns, though the report notes that most observed intrusions stem from the server‑side exploits. Palo Alto Networks emphasizes that its Advanced WildFire, URL Filtering, DNS Security, IP Defense, Cortex XDR and XSIAM solutions are engineered to detect and block the described techniques, and it offers incident‑response and intelligence services through Unit 42 for any organization suspecting compromise.

The actors complemented their primary infection chain with a phishing layer that relied on counterfeit PDFs and look‑alike domain names to lure victims. Researchers observed that the malicious sites were crafted to resemble official government portals, and the downloaded files were presented with a PDF icon even though they were executable binaries. When a user opened the “PDF,” the malware launched a genuine‑looking PDF in the background to preserve the illusion while the malicious code executed. One of the decoy documents mimicked a 2025 official release from the Union Election Commission of the Republic of the Union of Myanmar, underscoring the campaign’s focus on high‑priority, region‑specific targets.

Both PerfectPioneer RAT and the later KUKURAT family use HTTP/3 as their default command‑and‑control (C2) transport, a rarity in the broader malware ecosystem. The researchers explained that HTTP/3 runs over QUIC on UDP, delivering a fast, encrypted channel while sidestepping many traditional defenses. Because enterprise firewalls and IDS/IPS solutions are often tuned for TCP‑based traffic, outbound UDP 443 streams can slip past deep‑packet inspection, preventing passive TLS interception without breaking the connection.

The PerfectPioneer RAT, first seen in the wild in July 2025, is a custom Go‑based backdoor that has been iterated at least six times. Its deployment is highly versatile: the payload can be dropped as a standalone executable, delivered through DLL side‑loading, hidden inside decoy software installers, or spread via the phishing scheme described above. In the side‑loading scenario, the attackers altered a benign executable to embed XOR‑encrypted shellcode and paired it with a malicious loader DLL; when the host program runs, the DLL decrypts the shellcode and injects the RAT into memory. A separate campaign used a counterfeit installer that pretended to install Palo Alto Networks’ GlobalProtect client, while silently writing and launching the PerfectPioneer binary in the background.

The most feature‑rich sample, labeled “newhttp3,” was found in the path D:/proj/http3/newhttp3/agent/ and includes a broad command set compiled directly into the Go binary. Capabilities span full file‑system manipulation (list, upload, download, delete, chunked transfers, temporary‑file cleanup), shell command execution, process control (listing, killing, starting, COM‑based creation), classic code‑injection techniques (VirtualAllocEx, WriteProcessMemory, CreateRemoteThread), keylogging, a SOCKS5 proxy, screenshot capture, window enumeration and manipulation, and resource monitoring. Communication relies on HTTP/3 with an automatic fallback to HTTP/2, contacting four endpoints—/agent/register, /agent/heartbeat, /agent/tasks, and /agent/result—that later appear in KUKURAT traffic as well. To preserve connectivity, the RAT runs a self‑healing routine that watches an internal “connected” flag; if the flag flips false, the malware aborts current heartbeat and polling loops and forces a fresh registration with the C2 server.

The malware’s cleanup routine is engineered to erase its forensic trail and keep resource consumption low. It continuously watches for any temporary files generated during a session and deletes them on the fly, limiting persistent disk artifacts. At the same time, the RAT monitors its own memory use; if the process exceeds a 100 MB threshold it forces Go’s runtime garbage collector to run aggressively, purging unallocated memory and keeping CPU load modest. A parallel resource‑monitoring routine records the infected host’s CPU and memory usage every 30 seconds, storing a performance snapshot that the remote operator can request later with a dedicated command.

Analysts identified six distinct PerfectPioneer RAT variants, each compiled from the same codebase but differing in communication protocols and feature sets. The compiled project paths—such as D:/proj/RatAAkk-tt/agent/, D:/proj/PerfectRat/agent/, D:/proj/http3/newhttp3/agent/, D:/proj/http3/rat_http3_ls/agent/, D:/proj/http3/Pioneer/, and D:/proj/http3/rat_http3_cc/—show a clear evolution toward HTTP/3. Compilation timestamps were either deliberately altered or stripped, preventing a reliable build chronology. Some variants retain a minimal, lightweight footprint, while others bundle extensive credential‑harvesting, process‑manipulation, and tunneling capabilities; a full feature matrix is documented in the report’s Appendix A.

A notable safeguard, dubbed isDangerousCommand, appears in the Pioneer and rat_http3_cc variants. Before executing any inbound instruction, the routine checks the command against a hard‑coded blocklist of administrative actions that would generate noisy forensic evidence. “The blocklist includes commands that are highly destructive or forensics‑heavy,” the researchers noted, emphasizing that the list contains syntax for both Windows and Linux environments. Because Go code can be cross‑compiled, the presence of Linux‑style commands suggests the actors were simultaneously developing a Linux counterpart.

In March 2026 the group released a new payload named KUKURAT, written in C++ and deployed alongside existing PerfectPioneer infections. Two KUKURAT families have been observed: one that communicates over HTTP/2 and another that uses HTTP/3. Unlike its predecessor, KUKURAT is a fully modular framework; at least twelve distinct DLL modules can be fetched from the C2 server and loaded into memory only when needed, allowing the core binary to evade static scans. Upon launch the backdoor contacts a /register endpoint, posting hostname, username, integrity level, OS type and architecture, IP address, and a randomly generated agent instance ID. The server replies with a JSON configuration that includes a session ID, heuristic camouflage flags (network, registry, file all set to 0), and runtime parameters such as a 300 000 ms heartbeat timeout, 5 worker threads, and a 600 000 ms module execution timeout, among other retry and delay settings.

The agent periodically contacts a “/heartbeat” endpoint, sending a JSON payload that confirms it is still active and retrieves any pending commands; the full structure of this heartbeat message is reproduced in the report’s Appendix B.

KUKURAT’s core binary is deliberately lightweight, delegating post‑exploitation functions to plug‑in modules. Analysis of compiled strings reveals that the available modules support screenshot capture, keylogging, host and port scanning, process enumeration, termination and injection, file upload/download/list/delete/execute, native shell execution via cmd.exe and PowerShell, lateral movement through PsExec, SMB credential brute‑forcing, SOCKS tunneling, local privilege escalation, execution of Beacon Object Files, and a silent‑mode setting. The framework also permits loading arbitrary external modules that are not pre‑coded, giving the operators the ability to extend functionality on the fly.

During the investigation two additional executables were observed being dropped by KUKURAT: PsExecService.exe and SecurityGuard.exe. While the latter could not be recovered, PsExecService.exe was obtained and its PDB strings label it as a “KUKURAT external module” compiled as a standalone program. The tool appears to be a custom remote‑execution service derived in part from the open‑source MyPsExec project; when run it registers a Windows service named PsExec and creates a named pipe \\.\pipe\PSEXEC. The service accepts four commands—shell execution, file upload, file download, and directory listing—and automatically removes itself and cleans up after five minutes of inactivity to reduce forensic traces.

KUKURAT also embeds a heuristic camouflage engine that can fabricate activity across network, registry and file system dimensions. Via the C2 command set_obfuscation, the operator can enable or disable each vector. When active, the engine issues ten fake HTTP requests per cycle to random domains drawn from a hard‑coded list of fifteen popular sites, writes random data to the Chrome crash‑recovery registry key HKCU\Software\Google\Chrome\BrowserCrashRecovery and then reads and deletes it, enumerates the Chrome update client‑state key, and continuously creates and deletes temporary files with random names and contents in %temp%. This multi‑axis noise is intended to mask the malware’s true behavior from automated detection and forensic analysis.

A companion watchdog executable is used to keep KUKURAT running. The main binary spawns a thread that writes its own file path to a local named pipe called WINDOW_SEC_WATCHDOG every three seconds; the watchdog reads this path and polls a heartbeat every second. If no heartbeat arrives for fifteen seconds, the watchdog relaunches the agent using the cached path. In one case the watchdog was placed in C:\ProgramData\Cyvera\Ransomware\, a directory historically associated with Cortex XDR, suggesting an attempt to blend in with legitimate security‑product files.

Two variants of KUKURAT were identified, one using HTTP/2 and the other HTTP/3 for command‑and‑control traffic. Apart from the transport protocol, the codebases are nearly identical, but the HTTP/3 version replaces a monolithic nested if‑else command handler with an array of function pointers that map each command to its own routine. This refactor simplifies adding or removing capabilities and improves the malware’s adaptability and survivability.

Native exec command. The HTTP/3 build of KUKURAT adds a hard‑coded “exec” instruction that calls CreateProcessA to launch a subprocess on the infected host and streams the command‑line output back to the C2 server. Analysts note that this capability is absent from the HTTP/2 variant, indicating the developers wanted a built‑in way to run system commands without pulling additional DLLs from a staging host.

Lateral spread via log‑on scripts. After compromising a domain controller, the operators dropped malicious log‑on scripts that execute each time a legitimate user authenticates. The scripts silently download and run KUKURAT on the user’s workstation, leveraging familiar tools such as PsExec, WMI and WinRM for further movement. Cortex XDR alerts captured the unusual startup process running under a local or domain user SID, which investigators flagged as a sign of the log‑on script persistence technique.

Credential‑theft arsenal. The threat group harvested passwords through several avenues. They performed LSASS memory dumps by abusing the native comsvcs.dll utility with a command that invokes the MiniDump function by its hidden ordinal #+000024, a trick designed to evade detections that look for the “MiniDump” string. In parallel, they ran Mimikatz, ProcDump, SharpDPAPI and manual SAM/SYSTEM hive extraction to collect local hashes. To compromise the Active Directory itself, the actors executed DCSync attacks, pulling password hashes directly from the domain controller as shown in the campaign’s forensic traces.

Document exfiltration and mailbox takeover. Using the PerfectPioneer RAT alongside KUKURAT, the attackers exfiltrated large volumes of .pdf, .docx, .xlsx and .png files that appeared to contain sensitive material on international relations and internal politics. From a seized Exchange server they ran PowerShell commands to enumerate accounts with the ApplicationImpersonation role, then, when needed, granted a compromised account full mailbox permissions across the organization with Add‑MailboxPermission ‑AccessRights FullAccess, effectively giving the actors unrestricted read/write access to every mailbox.

The attackers leveraged a hidden PowerShell instance—C:\Windows\System32\WindowsPowerShell\v1.0\powershell.exe launched with -WindowStyle Hidden -NoLogo -NoProfile—to run a one‑liner that enumerated every mailbox (Get‑Mailbox -ResultSize Unlimited) and then piped each result to Add‑MailboxPermission. The command added the attacker‑controlled account (redacted in the report) as a FullAccess delegate on each primary SMTP address, explicitly disabling AutoMapping and suppressing confirmation prompts. Cortex XDR logs captured this activity, illustrated in the report’s Figure 8, confirming that the malicious web shell had first been used to map the Exchange environment before the bulk permission grant.

The operation, labeled TGR‑STA‑0049, employed the KUKURAT and PerfectPioneer RAT families, both of which now communicate over HTTP/3 via QUIC on UDP to evade traditional TCP‑centric monitoring. Analysts noted that PerfectPioneer’s self‑healing connection logic and KUKURAT’s heuristic camouflage engine allow the malware to persist while remaining invisible to signature‑based defenses. The binaries also contain hard‑coded safety switches that block “loud” or destructive commands, a design choice intended to keep the actors’ activity under the radar while they harvest what appear to be intelligence‑related files and abuse Exchange impersonation rights.

Defenders are urged to move beyond legacy visibility by inspecting or outright blocking outbound UDP 443 traffic, forcing any QUIC‑based sessions to fall back to TCP‑based HTTP/2 or TLS where deep‑packet inspection can operate. Additional hardening steps include continuous auditing of Sysvol and Group Policy Objects for unauthorized changes, creating alerts for rundll32.exe invoking comsvcs.dll with ordinal‑based calls (e.g., #000024) that bypass string filters, and regularly validating Exchange accounts for the ApplicationImpersonation role. Automated scripts that detect bulk Add‑MailboxPermission executions—especially those granting FullAccess without mapping—should be deployed to flag suspicious privilege escalation across the mail environment.

Palo Alto Networks responded by updating its WildFire machine‑learning models and reinforcing its Advanced URL Filtering, DNS Security, and IP Defense services to flag the domains and URLs tied to the campaign. Cortex XDR and XSIAM now leverage Behavioral Threat Protection and local analysis to block both known and novel malware variants. The Idira suite—Endpoint Privilege Manager, Privileged Access Management, and Secure Infrastructure Access—provides just‑in‑time elevation, zero standing privileges, and continuous session recording to curb risky actions. Unit 42 offers tailored threat intel, analyst consultations, and exposure tracking, with incident response contacts listed for each region. The findings have been shared with Cyber Threat Alliance members to accelerate protective deployments.

Indicators of Compromise – SHA256 hashes

  • PerfectPioneer RAT – RatAAkk‑tt:

  • 2281e754f70e11298db1a0a7c1b527cf2b5eec2056d4c73a46ed572ab873dc6e

  • 9721d65ee7695db2b3dc3705f5d8e5efe3a478c3b6d1704d66daf77fe95aeb0d

  • d725ad8b523c94f99fca9e72a650e835dca47b9ebe7610d6c4eb3525d410c365

  • PerfectPioneer RAT – PerfectRAT: (hashes not disclosed in the excerpt)

The forensic dump contains a series of SHA‑256 hashes that correspond to the various binaries used by the PerfectPioneer remote‑access tool. Among them are the primary “rat_http3_ls” payload (f4c9d5d13c1a09533876a02f6fa04831efbf600b05cfb3ed765a0ebde82b0457), its decoy installer package (92fb9545ea49e2fb4635fce1859ff3f609738be75f85d146614243fc8bc80b67), the core agent binaries (ab06f540ecdd1d45491aa71bfc04c80852371c8f5cd43b9a6ebb53f1e1b69588), and the modified host binary for the newhttp3 variant (8bf9401990d3255d1022a25884b02ee75ef01bd1a729d29ff52de2e5f0d637a2). Additional hashes identify the newhttp3 loader DLL (5f857ff78bfbef2a55a846fc93744e11e07a9b72ee3fc3bc345c32ce9615ddee), its core agents (61f7e2f4159c566578e88cb05cdf49e4378f45ebddbd332224e97c11df922120), and the rat_http3_cc modified host binary (a6a9ccffadd00b0e75e913b0f851b1f834fd20588df8b89759d9ae49a48c8e10) together with its loader DLL (b975f5932c04b381f8fa58a3303194f1aff950d3c6cd12dcc1d323bb3393718b) and core agents (8cfe629fc08ae06c8ba106113ba30bb16fa2fbbf8bd3dd6831d16b031e0ebe70). A further set of hashes is linked to an unknown PerfectPioneer variant and its modified host binary (0533365accbba405757de7eb486533c11c8e0510a28d414ee8c5e6b8a41ed97b). The package also includes hashes for generic loader DLL components (0c49e1b98a236d3391740473101b66dd48c719fa4112dbef3816b56bce58e6e7) and assorted supporting files.

The KUKURAT family is represented by separate HTTP/3 and HTTP/2 variants, identified by the hashes b8dca1c7e7a74ce82ab4e4cad9b718a458735d2c6c956e4cf3109aac59bfd0e4 and 9fb0976fcd1f92c4711726a91cff8be9a39944efedc4caf2d80f935327f8f490 respectively. A companion watchdog process (b062b4d8c19baf5c1568e28b3e0a5e2318e9501ea61ddfe804d9755c1bdbf887) and a PsExecService.exe module (34cc9bc4a533b91e7cdb30fecc57ba609dc890780fd13ecfbc253047b3b37b78) are also present, alongside the SharpDPAPI utility (0cbde9b7f59dafca80f214121a04eda7804b920da28ce1c06939a4783e6b097f). Several files could not be mapped to known binaries (e.g., 82a989c728db4e09a2b85a4a55f7663cefa62844632cb16dcda5ca70bf09a8a5) and a collection of web‑shell artifacts is listed under hashes such as 99f571c07170250a84f0ae9de4179044df91244855d9b39978729933ea3d438f.

Command‑and‑control infrastructure for the two families is enumerated in the dump. PerfectPioneer’s domains include www.maui‑cocktailbar[.]com, ads‑edge[.]com, www.wkuyda[.]com, www.thailisting[.]net, and www.yangon.it[.]com. KUKURAT’s servers are listed as www.cloudflacecdn[.]com, www.police.gr[.]com, and www.investigationj6[.]com. Two IPv4 addresses—27.254.164.212 and 27.254.164.209—appear as additional endpoints.

The “newhttp3” variant of PerfectPioneer implements a rich command set, as shown in the extracted table. It supports file‑system operations (file_info, file_list, file_delete, file_upload, file_download, file_download_chunk, file_upload_status), shell execution (shell_execute), process management (process_info, process_list, process_kill, process_kill_by_name, process_start, process_create_com, process_inject_shellcode), window manipulation (window_list, window_close, window_minimize), and key‑logger control (keylogger_start, keylogger_stop, keylogger_status). Each command is designed to be invoked over the HTTP/3 channel, enabling granular post‑exploitation control.

The analysis package also references supplemental material, including a “Check‑point” guide on QUIC, a technical overview of the Go garbage collector, and a Medium article discussing HTTP/3 security challenges. An Appendix (A) provides a comparative matrix of PerfectPioneer’s architectural evolution, highlighting which post‑exploitation capabilities and communication protocols have been added or removed across its development lifecycle.

The new HTTP/3‑enabled PerfectPioneer RAT includes a suite of control commands such as keylogger_get_events to pull logged keystrokes, keylogger_clear to empty the buffer, screenshot_capture and screenshot_cleanup for screen grabs, resource_stats and resource_history to report CPU and memory usage, socks5_proxy for proxy management, and a full update workflow ( update_check, update_start, update_status, update_cancel ).

The earliest known family, labeled RatAAkk‑tt, resides in D:/proj/RatAAkk-tt/agent/. It communicates exclusively over HTTP/2 and omits persistence, process‑injection, resource‑monitoring, and self‑update capabilities. Its only notable addition is a component called TerminalManagerV2, which provides a persistent interactive shell for the operator.

The PerfectRat variant (project path D:/proj/PerfectRat/agent/) also uses HTTP/2 and is delivered as a PDF‑styled payload, likely via phishing. It adds token‑impersonation for local privilege escalation, encryption modules, and a full remote‑desktop interface that lets the attacker view and control the victim’s mouse and keyboard. Like RatAAkk‑tt, it lacks persistence, injection, and SOCKS5 proxy functions.

Two later families shift to HTTP/3. The rat_http3_ls variant ( D:/proj/http3/rat_http3_ls/agent/) and the Pioneer variant ( D:/proj/http3/Pioneer/) are both PDF‑masqueraded and stripped of secondary post‑exploitation tools such as keyloggers, screenshot capture, SOCKS5 proxies, process injection and persistence, making them lightweight footholds for initial access. Pioneer introduces a protective routine named isDangerousCommand that blocks destructive actions. A subsequent branch, rat_http3_cc ( D:/proj/http3/rat_http3_cc/), builds on Pioneer by adding local persistence and embedding a tiny HTTP server listening on port 56793. The server exposes four endpoints: “/” for API discovery, “/info” for agent metadata, “/health” for a timestamped liveness check, and “/status” which returns a JSON flag indicating C2 connectivity.

The isDangerousCommand mechanism enforces three layers of defense. The first hard‑codes 41 keywords that would format disks, recursively delete files, shut down or reboot the host, launch fork‑bombs, alter permissions, manage users, kill services, reset firewalls, or tamper with Windows registry and boot settings. The second layer heuristically blocks 16 patterns associated with chained deletions, raw device writes, download‑and‑execute pipelines, Base64‑encoded PowerShell payloads, and silent echo commands. The final layer prevents any file‑removal command targeting twelve critical system paths (e.g., /etc/, /bin/, /c:/windows, c:/program files).

PerfectPioneer RAT achieves long‑term persistence by writing its executable path to the Windows Run key at HKCU\Software\Microsoft\Windows\CurrentVersion\Run. Investigators have not yet disclosed how widely these variants have been deployed or whether additional evasion techniques are in use, leaving the full scope of the campaign open for further analysis.

Why It Matters for Trust & Control Assurance — The use of QUIC highlights a gap in traditional network‑traffic monitoring controls; organizations must ensure detection capabilities cover UDP‑based protocols to maintain a defensible audit posture.

Who Is Affected — Government agencies in Southeast Asia; any organization that relies on legacy network security appliances not tuned for UDP/QUIC traffic.

Recommended Actions –

  • Extend network monitoring to include UDP‑based traffic and profile QUIC flows for anomalies.
  • Validate that endpoint detection and response (EDR) solutions can inspect encrypted HTTP/3 payloads or generate telemetry for downstream analysis.
  • Review and harden web‑server patching processes to eliminate exploitation pathways.

Source: https://unit42.paloaltonetworks.com/custom-rats-target-southeast-asia-governments/

Technical Notes —

  • Attack vector: Vulnerability exploitation of Microsoft Exchange/web servers and phishing with malicious PDFs.
  • Malware capabilities: Full file system access, process injection, keylogging, SOCKS5 proxy, synthetic activity generation, and an operational safety switch.
  • Protocol: HTTP/3 over QUIC (UDP 443), with fallback to HTTP/2.

Source: https://unit42.paloaltonetworks.com/custom-rats-target-southeast-asia-governments/

📰 Original Source
https://unit42.paloaltonetworks.com/custom-rats-target-southeast-asia-governments/ ↗

This Verisq Intelligence Brief is an independent analysis. Read the original reporting at the link above.

Third-party risk

Does this breach reach you?

Verisq continuously monitors your vendors for breach and ransomware activity, so the question stops being whether it happened and becomes whether it reaches you.

See a live Trust Center →