What happened
In September 2016, a botnet built from compromised routers, security cameras and digital video recorders began launching some of the largest distributed denial-of-service attacks recorded to that point, including floods directed at a security researcher's website and a French hosting provider. A US government alert issued on 14 October 2016 described the malware, later known as Mirai, as continuously scanning the internet for vulnerable internet-of-things devices and attempting to log in using a list of just 62 common factory-default username and password combinations, with more than 380,000 devices reportedly compromised across the campaigns the alert describes. An independent seven-month measurement study presented at the USENIX Security Symposium tracked the botnet to a peak of roughly 600,000 infections and found it spread almost entirely through embedded and IoT devices rather than conventional computers.
Confidence and limits
The government alert is a defensive advisory built from observed attack traffic and reported credential lists, and it is specific about the scanning method and device types affected. The USENIX study is an independent academic measurement based on the researchers' own network telemetry over a seven-month tracking window, giving the two documents different but overlapping vantage points on the botnet's scale. Neither document reviewed for this article addresses any individual's identity, prosecution or legal outcome.
Why it mattered
Mirai demonstrated that a botnet did not need to compromise sophisticated targets to cause internet-scale disruption; it needed only enough low-cost, rarely updated consumer and embedded devices left on factory-default credentials. The USENIX study's description of the attack as relying on novice malicious techniques against low-end devices to threaten even some of the best-defended targets captured why the case became a reference point for arguing that the security of the least-defended devices on a network can determine the resilience of the internet as a whole, not just of the device owner.
Defensive takeaway
Change every default credential on internet-connected embedded devices before they are deployed, disable remote administration features you do not use, and keep device firmware current, since these devices are rarely monitored the way servers are.
- Do you have a complete inventory of internet-connected cameras, routers, recorders and other embedded devices on your network?
- Have factory-default administrative credentials been changed on every one of those devices, not just the ones you remember installing?
- Would unusual outbound scanning traffic from an embedded device be detected by your monitoring, or only traffic from servers and workstations?
The defensive lesson from both documents is narrower and more durable than any single attack they describe: a very large share of internet disruption capacity can come from devices nobody thinks of as computers, and default credentials left unchanged at the factory remain, years later, one of the most reliably exploitable weaknesses an organization can control.