Skip to content
NewCybercrime Report 2026 · Eight Chapters · A Post Is Not An Incident → Explore The ChaptersCybercrime Report 2026 → Explore

Odin Research / What is / Malware and backdoors

What Is the Mirai Botnet? How Mirai Infects IoT Devices

The IoT botnet that broke DDoS records in 2016 and never really left. How it spreads, why its source code still matters, and what infected routers look like today.

By
Odin by Brandefense
Updated
21 Nov 2026
Reading time
7 min read
Type
IoT botnet

In short

The Mirai botnet is malware that infects internet-connected devices such as home routers and IP cameras, usually by logging in with factory-default passwords. Infected devices join a botnet that launches distributed denial-of-service (DDoS) attacks. Mirai's source code was published in 2016, so many variants still spread today, often with added exploits.

Key takeaways

What it is
IoT malware that turns routers, cameras and similar devices into a DDoS botnet. Its source code went public at the end of September 2016.
How it works
It scans the internet for devices with Telnet open, logs in with a short list of default credentials and loads the bot into memory.
Why it matters
Mirai launched record DDoS attacks in 2016. Variants keep adding exploits for new device types, and infected routers also serve as proxy hops.
What Odin Research saw
353 Mirai endpoints on 279 hosts in Monograph 19, mostly consumer routers in China, Taiwan, Ethiopia and the Philippines. The owners are victims.
What to do first
Change default passwords, disable Telnet and UPnP where not needed, patch device firmware, and do not permanently block infected home routers.

What is the Mirai botnet?

The Mirai botnet is malware that infects internet-connected devices, mostly home routers and IP cameras, and turns them into a botnet. It usually gets in by logging in with factory-default usernames and passwords. The infected devices then launch distributed denial-of-service (DDoS) attacks on command.

Mirai became famous in late 2016. CISA's alert TA16-288A describes a DDoS attack of more than 620 Gbps on the Krebs on Security website on 20 September 2016. A few days later, an attack on the hosting provider OVH reached at least 1.1 Tbps, a record at the time.

Mirai's source code was published online at the end of September 2016, according to CISA. That release is why "Mirai" today means a large family of variants, not one botnet run by one group.

The device owners are victims. Most never notice the infection, because the device keeps working.

How the Mirai botnet works

Mirai's design is simple, which is part of why it spread so far.

  1. Scan. Infected devices scan the internet for other devices with weak security. CISA advises monitoring TCP ports 23 and 2323 for Telnet attempts.
  2. Log in. The bot tries a short list of common default credentials. CISA counted 62 username and password pairs. The research team behind Cloudflare's retrospective counted 64 in the initial version.
  3. Load. The bot runs from memory. CISA notes that a reboot clears it, but the device is reinfected quickly if the password stays the same.
  4. Attack. Operators send commands, and every bot floods the chosen target at once.

That is brute force in its crudest form: no exploit needed when the password is printed in the manual. MITRE ATT&CK files the end goal under Network Denial of Service (T1498).

Later variants added exploits. CISA described a variant in late November 2016 that scanned TCP port 7547 on broadband routers. Palo Alto Networks Unit 42 reported in June 2019 a variant carrying 18 exploits, 8 of them new to Mirai. It targeted wireless presentation systems, set-top boxes, SD-WAN devices, smart home controllers, IP cameras, NAS devices and routers.

Mirai botnet history and scale

A team from Cloudflare, Georgia Tech, Google, Akamai and several universities measured Mirai in a paper for USENIX Security 2017. Cloudflare published a summary by Elie Bursztein in December 2017.

  • Size. The researchers measured a peak of more than 600,000 infected devices, mostly home routers and cameras.
  • Dyn, 21 October 2016. The researchers studied the attack on the DNS provider Dyn. They concluded it was linked to attacks on gaming platforms rather than an effort to take down the internet.
  • Deutsche Telekom, 26 November 2016. A variant's router exploit caused outages affecting about 900,000 routers.
  • Liberia. The mobile operator Lonestar Cell was attacked 616 times from 31 October 2016.

Nearly a decade later, the code base still drives new botnets. Many families, such as Moobot, are built on it.

Why the Mirai botnet still matters to defenders

Mirai matters for two reasons today.

First, DDoS. Any organization with an internet-facing service can become a target, and Mirai-style botnets are cheap to rent or build.

Second, infected routers are infrastructure for other actors. Monograph 19 notes that infected routers are often reused as proxy or scanning hops in wider internet activity. A login attempt or scan from a home router may be a Mirai bot relaying someone else's traffic.

That creates a blocking problem. Home broadband addresses change hands often. A permanent block on an infected router's address mostly hurts the next customer who receives it. Treat Mirai hits as short-lived signals, not as permanent indicators.

What Odin Research observed

  • 353 Mirai endpoints. Monograph 19 counted 353 Mirai endpoints on 279 hosts, part of 618 infected IoT endpoints alongside Mozi (263) and Moobot (2). Of the Mirai endpoints, 63 are protocol-confirmed infections.
  • Consumer networks. China led with 107 endpoints on China Unicom and Chinanet. Taiwan followed with 53, Ethiopia with 43, the Philippines with 30 and Albania with 28.
  • Router defaults. Many devices presented default router certificates. Nineteen Taiwanese routers shared one TLS fingerprint, which points to the same device model rather than the same owner.
  • No link to the C2 servers. The routers shared no address, certificate or observed traffic with the C2 and password-cracking servers in the same report. Ten sat on dynamic telecom lines, where addresses change hands quickly.
  • Earlier and later sightings. In Monograph 02, Mirai made up 112 of 203 IoT botnet records. In Monograph 29, two Moobot nodes, a Mirai variant, served payloads over plain HTTP on port 80.

Odin Research's verdict is the same in every report: these routers are victims. They belong on short watchlists, never on permanent block lists.

How to detect the Mirai botnet

  • Telnet from your devices. Outbound connections from cameras, DVRs or routers to many addresses on TCP 23 or 2323 suggest an infected device scanning for others.
  • Inbound Telnet bursts. Repeated Telnet login attempts with default credentials against your edge show Mirai-style scanning.
  • Unexpected traffic spikes. An IoT device suddenly sending large volumes of UDP or SYN traffic may be taking part in an attack.
  • Plain HTTP payload fetches. Moobot stages payloads over HTTP. Monograph 29 publishes a Suricata signature (SID 9529005) for a /moobot. fetch with a Wget/ user agent, available in the Defense Kit.

For your own perimeter, an inventory of exposed devices matters most. Our guide to unknown exposed assets covers how to find them.

How to defend against and remove Mirai

CISA's guidance remains the baseline.

  1. Remove an infection. Disconnect the device, reboot it while offline, change the default password to a strong one, then reconnect.
  2. Change all default passwords before a device goes online.
  3. Patch firmware as soon as updates are available. Variants rely on known device flaws; see patch management.
  4. Disable UPnP and Telnet on routers unless you need them.
  5. Segment IoT devices. Put cameras and similar devices on their own network segment with no route to business systems.
  6. Buy from vendors that patch. CISA advises choosing IoT vendors with a reputation for security.

Mirai botnet compared with other IoT and edge botnets

BotnetHow it spreadsWhat Odin Research saw
MiraiDefault Telnet credentials, plus exploits in variants353 endpoints, 63 confirmed infections (Monograph 19)
MoziPeer-to-peer mesh of consumer gateways263 endpoints on 261 hosts (Monograph 19)
MoobotMirai variant that stages payloads over HTTP2 candidates in Monograph 19; 2 HTTP nodes in Monograph 29
P2PInfectPeer-to-peer worm through exposed Redis and SSH servers16 nodes, mostly on Alibaba Cloud (Monograph 29)

Frequently asked questions about Mirai botnet

What is the Mirai botnet?

Mirai is malware that infects IoT devices such as home routers and IP cameras, usually through default passwords, and uses them to launch distributed denial-of-service attacks.

Is Mirai still active?

Yes, as a family. The original source code was published in 2016, and variants keep appearing. Odin Research counted 353 Mirai endpoints in a two-week window ending 18 November 2026.

How does Mirai infect devices?

It scans the internet for devices with Telnet open and tries a short list of default usernames and passwords. CISA counted 62 such pairs. Later variants also use exploits for specific device flaws.

Who created Mirai?

Mirai's original author published the source code on a hacking forum at the end of September 2016, according to CISA and the Cloudflare retrospective. Since then, many unrelated groups have built their own variants.

How do I remove Mirai from a device?

CISA advises disconnecting the device, rebooting it while offline, changing the default password to a strong one and only then reconnecting. Mirai runs in memory, so a reboot clears it, but it returns if the password stays the same.

How do I know if my router is part of a botnet?

Warning signs include outbound Telnet scanning to many addresses, unexplained traffic spikes and slow performance. Check for firmware updates, change the admin password and disable remote management you do not use.

Should I block IP addresses of Mirai-infected routers?

Not permanently. Odin Research found most of them on home broadband, where addresses change hands. Use short watchlists so you do not block the next, uninfected customer.

Mirai vs Mozi: what is the difference?

Both infect consumer routers. Mirai is the older family whose published source code spawned many variants. Mozi organizes infected devices into a peer-to-peer mesh, which Odin Research observed on 263 endpoints.

Sources

  1. CISA, Alert TA16-288A, "Heightened DDoS Threat Posed by Mirai and Other Botnets". www.cisa.gov/news-events/alerts/2016/10/14/heightened-ddos-threat-posed-mirai-and-other-botnets (checked 21 November 2026)
  2. Cloudflare (Elie Bursztein), "Inside the infamous Mirai IoT Botnet: A Retrospective Analysis", December 2017. blog.cloudflare.com/inside-mirai-the-infamous-iot-botnet-a-retrospective-analysis (checked 21 November 2026)
  3. Palo Alto Networks Unit 42, "New Mirai Variant Adds 8 New Exploits, Targets Additional IoT Devices", 6 June 2019. unit42.paloaltonetworks.com/new-mirai-variant-adds-8-new-exploits-targets-additional-iot-devices/ (checked 21 November 2026)
  4. MITRE ATT&CK, Network Denial of Service, T1498. attack.mitre.org/techniques/T1498/ (checked 21 November 2026)

Related tools and terms

All Odin Research explanations · Brandefense glossary

Odin Research

Get the rules, not just the definition

Every Odin Research report ships with verified indicators and detection rules you can load today.

Open the Defense Kit