The Asia Pacific Network Information Centre (APNIC) has completed a comprehensive modernization of its global honeynet infrastructure, offering a much wider lens into malicious automated internet activity. By upgrading its backend data-processing systems, refreshing core sensors like Cowrie and Dionaea, and integrating new monitoring tools such as OpenCanary and Webpot, APNIC has significantly expanded its visibility beyond traditional remote-access services. The newly integrated network now captures traffic targeting specialized vectors, including simulated Fortinet SSL-VPN portals and Android devices with exposed debugging ports.
Captured malware flows directly into an advanced Linux sandbox connected to major threat intelligence platforms, including MISP and Abuse.ch. While researchers emphasize that it remains too early for exhaustive long-term evaluations, the freshly deployed network architecture has already begun unmasking expansive patterns of automated malicious activity.
During a single monitoring window from September 21 to September 27, nearly 99,000 distinct IP addresses interacted with APNIC’s distributed honeypots, generating an astonishing 173 million logged events. Cybersecurity analysts measure these interactions primarily by unique source internet protocol addresses rather than raw event volume. Because a single automated botnet client can generate millions of individual log entries while another device connects only once, counting unique IP addresses provides a far more accurate representation of the sheer volume of distinct machines aggressively probing global sensors.
The scope of the observed threat landscape is heavily concentrated. Nearly three-quarters of all attacking entities targeted foundational remote administration protocols like SSH and Telnet. The Cowrie SSH and Telnet honeypot alone attracted 72,256 distinct IP addresses, accounting for 73 percent of all observed attacker traffic and generating 121.1 million separate events. In contrast, sensors emulating Windows and database services, such as Dionaea, logged 21,210 unique IP addresses, representing 21 percent of the total attacker pool across 12.9 million events.

Other segments of the honeynet observed substantial engagement as well. OpenCanary, which mimics various remote access and database interfaces, recorded 14,393 unique IPs and 34.5 million events, demonstrating a high frequency of persistent connection attempts per attacker. Specialized infrastructure emulations attracted smaller but notable shares of hostile traffic: the Fortinet SSL-VPN portal simulator caught 1,876 unique IPs generating 4.4 million events; ADBHoney, posing as an Android device with an open debug port, logged 1,841 IPs; and the experimental Webpot application trap recorded 1,723 distinct IP addresses.
An examination of geographic origins reveals that attacking infrastructure is split remarkably evenly between the Asia Pacific and European Regional Internet Registry regions, with Latin America following as a distant third. Analysts caution that this distribution highlights the physical locations of compromised consumer hardware, enterprise systems, and scanning infrastructure rather than the home nations of the primary threat actors. Because modern botnets routinely harness infected machines scattered across multiple continents, regional data serves primarily to help network administrators and national computer emergency response teams pinpoint where remediation efforts can achieve the greatest systemic impact.
When examining the specific credentials utilized in brute-force campaigns, researchers found that attackers consistently rely on lazy, highly predictable patterns, including factory defaults, short number sequences, and usernames mirrored as passwords. Across thousands of SSH and Telnet login attempts, standard credentials such as root, admin, support, ubuntu, and user dominated the traffic logs, paired with trivial passwords like 123456, password, 123, and blank entries. Similarly, Microsoft SQL Server attack telemetry showed that out of thousands of probing IP addresses attempting to access the built-in administrator account, the vast majority relied on blank passwords or generic dictionary words, heavily pointing to the widespread deployment of automated scripts and shared wordlists.
Digging deeper into specific honeypot findings, the data exposes the remarkable persistence of legacy malware strains that continue to plague modern digital infrastructure. Through the Cowrie SSH and Telnet sensors, researchers recorded thousands of unique download URLs and file-transfer attempts, uncovering hundreds of distinct malware binaries. The most widely distributed samples traced back to older IoT botnet families such as Mozi and Mirai. Despite previous international law enforcement actions and reported kill switches designed to neutralize them, these ghost botnets have shown persistent resilience, with recent spikes in activity heavily concentrated within specific economies in the Asia Pacific region.
Concurrently, SSH-focused sensors documented thousands of instances where attackers automatically deployed specific script one-liners designed to wipe existing SSH configuration directories, install unauthorized cryptographic keys, and lock down file permissions. The digital fingerprints of these keys frequently connect back to established cryptomining botnets such as Outlaw, which aggressively propagate across the internet by exploiting weak administrative passwords.

Legacy vulnerabilities remain a primary driver of automated exploitation across Windows and database environments monitored by the Dionaea honeypot. Server Message Block file-sharing protocols attracted massive interest, with thousands of hostile IPs actively pushing malware samples into the sensors. A significant portion of these widespread file deliveries consisted of components tied directly to WannaCry, the infamous ransomware worm that first wreaked global havoc through the EternalBlue exploit. The fact that thousands of machines continue to distribute these payloads highlights the reality that unpatched, legacy Windows systems remain active on the internet years after critical patches were issued.
Meanwhile, experimental traps like Webpot and specialized infrastructure emulators demonstrate that threat actors continuously test both old and emerging vulnerabilities. While long-standing flaws such as decades-old command injection vulnerabilities continue to draw heavy automated reconnaissance traffic, newly disclosed security gaps in enterprise applications and developer frameworks quickly join the standard automated scanning rotation. Similarly, honeypots designed to mimic FortiGate firewalls and exposed Android debugging interfaces revealed aggressive credential spraying and automated attempts by threat actors to install unauthorized cryptominers and eliminate competing botnet software.
Ultimately, the comprehensive telemetry gathered by APNIC’s rebuilt honeynet underscores an enduring reality of modern cybersecurity. The vast majority of malicious internet traffic is driven not by sophisticated zero-day exploits, but by the relentless abuse of legacy malware strains, default router credentials, and long-patched vulnerabilities that continue to compromise tens of thousands of unsecured machines every single week.