ASUS Router Botnet Attack: AI Uncovers Hidden Backdoor

Forecast = Mostly cloudy with a chance of rogue SSH access—keep your patches up to avoid a phishy forecast! Welcome to Storm⚡️Watch, where we unpack the latest in cybersecurity threats, research, and the tools that keep the digital world safe. In this episode, we invite GreyNoise Security Architect and researcher Matthew Remacle (a.k.a., Remy) to kick things off with a deep dive into a fascinating and highly sophisticated botnet campaign targeting ASUS routers—a story that starts with a little help from machine learning and ends with some hard lessons for defenders everywhere. GreyNoise researchers spotted this campaign using SIFT, their AI-powered network traffic analyzer, which sifted through more than 23 billion network entries and managed to flag just 30 suspicious payloads targeting ASUS routers. What made this botnet stand out was its surgical precision and stealth—far from the usual noisy, attention-grabbing attacks. The attackers knew exactly what they were doing, focusing on disabling TrendMicro security features embedded in the routers, essentially breaking in by first turning off the alarm. The attack chain reads like a masterclass in persistence: brute force and clever authentication bypasses got them in the door, a null byte injection tricked the router's authentication system, and a command injection vulnerability allowed them to manipulate logging features in a way that opened up even more attack paths. The real kicker? The final backdoor was installed using legitimate ASUS features, meaning it could survive firmware updates and stay hidden from traditional detection methods. This campaign affected thousands of routers globally, with over 4,800 compromised devices detected and counting. Even after ASUS released a patch—adding character validation rather than fixing the underlying flaw—researchers found that the fundamental vulnerability remained, and attackers could potentially work around the patch. This story highlights the ongoing challenges in IoT security: complexity breeds vulnerability, persistence is a nightmare to detect and remove when attackers use legitimate features, and patches often address symptoms rather than root causes. It's a reminder that traditional signature-based detection is no longer enough—behavioral analysis and AI-driven anomaly detection are now essential for spotting these advanced threats. We also touch on the bigger picture: the evolving cat-and-mouse game between attackers and defenders, the importance of defense in depth, and why understanding normal network behavior is more critical than ever. Plus, we look at the human element—attackers who are patient, technically sophisticated, and deeply aware of how to evade detection. For organizations, the takeaways are clear: defense in depth, behavioral monitoring, asset management, and patch management are all non-negotiable. And for everyone else, it's a reminder that the devices we trust to protect us are themselves complex and potentially vulnerable computers. Later in the episode, we take a closer look at vulnerability scoring systems—CVSS, EPSS, and SSVC—and why reading between the scores is so important for risk management. We also highlight the value of fresh, actionable data from sources like Censys and VulnCheck, and round things out with a nod to the ongoing conversation happening on the GreyNoise blog. Thanks for tuning in to Storm⚡️Watch. Stay vigilant, keep learning, and remember: in cybersecurity, the difference between safe and compromised can be as subtle as a single null byte. Storm Watch Homepage >> Learn more about GreyNoise >>