Compromised IoT Network Activity
Medium severity TLP:CLEAR Part 2 of 2 Research by ESET Research Published 21 Nov 2026
Hypothesis
End-of-life IoT devices have been conscripted into botnets or hijacked for traffic redirection, evidenced by anomalous DNS failure patterns, internal lateral probing, or HTTP connections to phishing infrastructure.
Brief
The Risk of Abandoned Hardware
End-of-life (EoL) IoT devices represent a significant blind spot in modern networks. As ESET Research notes in their article, My IoT devices are no longer supported – is my smart home safe?, these devices often continue functioning long after manufacturers stop providing security patches. Without updates, they become easy targets for botnet operators and traffic redirection schemes.
How the Hunt Flows
This hunt does not begin with broad, expensive scanning. Instead, it uses a gated flow to identify high-confidence leads before investigating specific behaviors.
First, a lead query examines the hb_dns_activity surface. The query seeks internal IP addresses generating a high volume of NXDOMAIN or SERVFAIL responses. This pattern often indicates a botnet agent attempting to reach command-and-control (C2) servers via Domain Generation Algorithms (DGA) where the domains have not yet been registered or have already been taken down.
Once an analyst confirms these DNS leads, the hunt gates the next phase. If the behavior looks automated rather than accidental, the hunt pivots into parallel behavioral analysis. It searches the hb_network_connection surface for internal management port probing—such as Telnet or ADB—directed at multiple internal targets. Simultaneously, it checks the hb_http_activity surface for connections to known phishing or redirection domains, which suggests the device is acting as a malicious gateway.
Finally, the analyst correlates these findings. A device showing both C2-style DNS failures and internal scanning is a high-confidence candidate for isolation.
What This Hunt Cannot See
This hunt focuses on the L3 gateway level. If an adversary moves laterally between two devices on the same L2 subnet without passing through a monitored router or switch, the network connection queries will not see the traffic. Additionally, if the botnet uses DNS-over-HTTPS (DoH) to hide its lookups, the lead DNS query will remain silent, forcing the hunt to rely entirely on the later behavioral steps. Identifying the specific physical device also requires accurate asset management or MAC OUI data, which may be missing in unmanaged environments.
Why This Is a Hunt, Not a Detection
While a single rule might catch a static malicious domain, this hunt identifies the pivot from C2 failures to internal lateral movement across multiple telemetry surfaces. It uses a gated flow to manage the cost of behavioral queries and requires an analyst to confirm the automated nature of the traffic before proceeding to containment.
In this series
Steps
-
Query, scoping
Lead: High DNS failure volume
Identify internal IPs generating a high volume of NXDOMAIN or SERVFAIL responses, suggesting automated C2 or DGA attempts.
sql, reads hb_dns_activitySELECT src_endpoint_ip, COUNT(*) as fail_count, COUNT(DISTINCT query_hostname) as unique_domains, MIN(time) as first_seen, MAX(time) as last_seenFROM hb_dns_activityWHERE rcode IN ('NXDOMAIN', 'SERVFAIL') AND time >= datetime('now', '-{{lookback_days}} days')GROUP BY src_endpoint_ipHAVING fail_count > 50ORDER BY fail_count DESCWhat a hit looks like. A list of IPs with persistent lookup failures. Silence suggests no heavy automated C2 activity is visible on this surface.
-
Agent triage
Evaluate DNS leads
Decide if the DNS failure pattern indicates a likely IoT botnet conscription.
-
Decision
Gate on suspicious DNS
Restrict expensive behavioral queries to only those environments showing suspicious DNS signals.
-
Query, detection candidate
Internal management port probing
Detect anomalous internal scanning of management ports originating from the suspicious IPs.
sql, reads hb_network_connectionSELECT src_endpoint_ip, dst_endpoint_port, COUNT(DISTINCT dst_endpoint_ip) as internal_targets, MIN(time) as first_seen, MAX(time) as last_seenFROM hb_network_connectionWHERE direction = 'outbound' AND ('{{scope_hosts}}' = '' OR instr(',' || '{{scope_hosts}}' || ',', ',' || src_endpoint_ip || ',') > 0) AND instr(',' || '{{management_ports}}' || ',', ',' || CAST(dst_endpoint_port AS TEXT) || ',') > 0 AND time >= datetime('now', '-{{lookback_days}} days')GROUP BY src_endpoint_ip, dst_endpoint_portHAVING internal_targets > 5ORDER BY internal_targets DESCWhat a hit looks like. One IP hitting multiple internal targets on ports like 23 (Telnet) or 5555 (ADB).
-
Query, enrichment
HTTP phishing traffic
Identify HTTP requests to known phishing domains, indicating a hijacked IoT gateway or router.
sql, reads hb_http_activitySELECT src_endpoint_ip, url_hostname, user_agent, COUNT(*) as req_count, MIN(time) as first_seenFROM hb_http_activityWHERE ('{{scope_hosts}}' = '' OR instr(',' || '{{scope_hosts}}' || ',', ',' || src_endpoint_ip || ',') > 0) AND instr(',' || '{{phishing_indicators}}' || ',', ',' || LOWER(url_hostname) || ',') > 0 AND time >= datetime('now', '-{{lookback_days}} days')GROUP BY src_endpoint_ip, url_hostname, user_agentORDER BY req_count DESCWhat a hit looks like. Connections to phishing infrastructure from internal IPs.
-
Agent triage
Confirm IoT compromise
Correlate all network-level findings to reach a final verdict on device compromise.
-
Decision
Route to containment
Initiate automated or manual containment for devices with a malicious verdict.
-
Response action
Isolate compromised IP
Contain the malicious IoT device at the network level.
-
Analyst task
Analyst validation
Verify the triage findings and identify the physical device.
-
Analyst task
Hunt close-out
Document results and plan for device decommissioning.
Coverage
3 of 5 stages covered
| Stage and ATT&CK technique | Covered | How, or why not |
|---|---|---|
| Botnet Conscription and C2 Communication T1190 |
Yes | dns-failed-lookups-lead |
| Traffic Redirection to Phishing Sites T1566 |
Yes | http-phishing-traffic |
| Lateral Pivoting from IoT Devices T1190 |
Yes | internal-port-probing |
| Exploitation of EoL IoT Devices T1190 |
Out of scope | Belongs to another part of the 'My IoT devices are no longer supported – is my smart home safe?' series. |
| Operational Impact on Smart Devices T1190 |
Out of scope | Belongs to another part of the 'My IoT devices are no longer supported – is my smart home safe?' series. |
Blind spots
- Needs hb_network_connection from access-layer switches. Lateral movement between devices on the same subnet that does not pass through a monitored L3 gateway will be invisible to these queries. It would answer: whether the device is communicating with other hosts on the same L2 subnet Remediation: Deploy network sensors at the access layer or use private VLANs to force all traffic through a monitored point.
- Needs accurate CMDB or MAC OUI integration. Without agent coverage, identifying the physical device for remediation is delayed, extending the window of risk. It would answer: what the specific model and owner are for the identified suspicious IP Remediation: Integrate DHCP fingerprinting or MAC OUI lookups into the central telemetry lake.
- Needs hb_http_activity targeting known DoH endpoints. If the botnet uses DoH, the lead query will be silent, and the compromise will only be visible in later behavioral stages. It would answer: whether the botnet is using DNS-over-HTTPS to hide C2 lookups Remediation: Block outbound DoH traffic from IoT subnets and enforce internal DNS resolution.
Parameters and data
Parameters
| Parameter | Type | Default | What it is |
|---|---|---|---|
lookback_days | number | 14 | Days of history to examine. |
management_ports | list[string] | 22, 23, 80, 443, 3389, 5555, 5900 | Management ports commonly targeted during internal propagation. |
phishing_indicators | list[domain] | login.router-update.invalid, security-check.wifi-auth.invalid, portal.isp-verify.invalid | Known malicious domains used for redirection or phishing. |
scope_hosts | list[host] | none | Suspicious IPs identified from the lead; leave empty to scan the entire estate in behavioral steps. |
Data sources
| Source | Category | Telemetry |
|---|---|---|
| Endpoint telemetry (hb_ surfaces) | endpoint | endpoint |
| Network telemetry | network | network |
| Web server / proxy logs | siem | network |
Source
An open hunt.md file. It runs anywhere that reads the format.
---
analysis: This hunt uses a gated flow to correlate DNS failure patterns with follow-on
behavioral evidence. A single rule would only catch a static domain; this hunt identifies
the pivot from C2 failure to internal lateral movement across multiple telemetry
surfaces.
blind_spots:
- id: gateway-logging-gap
owner: Network Engineering
question: whether the device is communicating with other hosts on the same L2 subnet
remediation: Deploy network sensors at the access layer or use private VLANs to
force all traffic through a monitored point.
requires: hb_network_connection from access-layer switches
risk: Lateral movement between devices on the same subnet that does not pass through
a monitored L3 gateway will be invisible to these queries.
stage: botnet-conscription-and-c2
- id: unmanaged-device-blind-spot
owner: Asset Management
question: what the specific model and owner are for the identified suspicious IP
remediation: Integrate DHCP fingerprinting or MAC OUI lookups into the central telemetry
lake.
requires: accurate CMDB or MAC OUI integration
risk: Without agent coverage, identifying the physical device for remediation is
delayed, extending the window of risk.
stage: lateral-movement-from-iot
- id: encrypted-dns-blind-spot
owner: Security Operations
question: whether the botnet is using DNS-over-HTTPS to hide C2 lookups
remediation: Block outbound DoH traffic from IoT subnets and enforce internal DNS
resolution.
requires: hb_http_activity targeting known DoH endpoints
risk: If the botnet uses DoH, the lead query will be silent, and the compromise
will only be visible in later behavioral stages.
stage: botnet-conscription-and-c2
coverage:
- stage: botnet-conscription-and-c2
status: covered
steps:
- dns-failed-lookups-lead
- stage: dns-redirection-and-phishing
status: covered
steps:
- http-phishing-traffic
- stage: lateral-movement-from-iot
status: covered
steps:
- internal-port-probing
- reason: "Belongs to another part of the 'My IoT devices are no longer supported\
\ \u2013 is my smart home safe?' series."
stage: initial-access-eol-exploitation
status: out_of_scope
- reason: "Belongs to another part of the 'My IoT devices are no longer supported\
\ \u2013 is my smart home safe?' series."
stage: operational-impact
status: out_of_scope
guardrails:
claims: no_unsupported
evidence: citation_required
missing_data: not_benign
telemetry: untrusted
hunt:
applicability: campaign-specific
handoff: promote-to-detection
justification: Unsupported and end-of-life IoT devices are vulnerable beachheads
that lack standard security controls; identifying their compromise through network
behavior is vital to protecting the primary enterprise network.
methodology: model-assisted
trigger: intel-report
hypothesis: End-of-life IoT devices have been conscripted into botnets or hijacked
for traffic redirection, evidenced by anomalous DNS failure patterns, internal lateral
probing, or HTTP connections to phishing infrastructure.
labels:
- hunt
- attack.t1190
- attack.t1566
- attack.t1071.001
- attack.t1071.004
- execution
- impact
- initial access
- lateral movement
name: Compromised IoT Network Activity
parameters:
lookback_days:
default: '14'
description: Days of history to examine.
from:
kind: manual
observed: '2024-01-01'
ref: hunt-standard
type: number
management_ports:
default:
- '22'
- '23'
- '80'
- '443'
- '3389'
- '5555'
- '5900'
description: Management ports commonly targeted during internal propagation.
from:
kind: manual
observed: '2024-01-01'
ref: threat-intel
type: list[string]
phishing_indicators:
default:
- login.router-update.invalid
- security-check.wifi-auth.invalid
- portal.isp-verify.invalid
description: Known malicious domains used for redirection or phishing.
from:
kind: manual
observed: '2024-01-01'
ref: threat-intel
type: list[domain]
scope_hosts:
default: []
description: Suspicious IPs identified from the lead; leave empty to scan the
entire estate in behavioral steps.
from:
kind: manual
observed: '2024-01-01'
ref: analyst-scoping
type: list[host]
provenance:
authors:
- name: Huntbase hunt generation
org: huntbase.io
generated:
by: huntbase-hunt-generation
from: https://www.welivesecurity.com/en/internet-of-things/iot-devices-no-longer-supported-smart-home-safe/
gates:
- dry-run
- lint
model: hb_google/gemini-3-flash-preview
rationale: Target IoT-specific VLANs, guest networks, and industrial control subnets
first. Focus the lead query on these zones to minimize noise from enterprise IT
equipment.
references:
- name: "ESET Research \u2014 IoT devices no longer supported"
url: https://www.welivesecurity.com/en/internet-of-things/iot-devices-no-longer-supported-smart-home-safe/
related:
- hunt: initial-access-eol-exploitation
reason: Vulnerability management and the exploitation of EoL software are handled
in the vulnerability-focused hunt in this series.
relation: out-of-scope-alternative
- hunt: exposed-iot-posture-operational-impact
relation: follows
scenario:
stages:
- name: Exploitation of EoL IoT Devices
observables:
- exposed management ports on DVRs and webcams
- unpatched firmware on refrigerator controllers
- no-name Android TV boxes with insecure services
- pre-installed malware on smart devices
slug: initial-access-eol-exploitation
tactic: initial-access
techniques:
- T1190
- name: Botnet Conscription and C2 Communication
observables:
- Mozi botnet communication
- BADBOX 2.0 botnet traffic
- outbound connections to known botnet controllers
- unusual process launches on Android TV boxes
slug: botnet-conscription-and-c2
tactic: execution
techniques:
- T1190
- name: Traffic Redirection to Phishing Sites
observables:
- router DNS settings redirection
- HTTP traffic to malicious phishing domains
- hijacked router management traffic
slug: dns-redirection-and-phishing
tactic: initial-access
techniques:
- T1566
- name: Lateral Pivoting from IoT Devices
observables:
- internal network scanning from IoT device IP addresses
- unauthorized connections to computer management ports from smart TVs
- accessing internal assets via hijacked routers
slug: lateral-movement-from-iot
tactic: lateral-movement
techniques:
- T1190
- name: Operational Impact on Smart Devices
observables:
- unauthorized temperature adjustments in smart controllers
- remote operation of smart door locks
- accessing unauthorized smart camera feeds
slug: operational-impact
tactic: impact
techniques:
- T1190
summary: Attackers target unpatched vulnerabilities in End-of-Life (EoL) IoT devices,
such as routers, cameras, and smart controllers, to gain initial network access.
Once compromised, these devices are used to pivot into internal networks, redirect
users to phishing sites, or are conscripted into large-scale botnets like Mozi
and BADBOX 2.0.
series:
index: 2
slug: my-iot-devices-are-no-longer-supported-is-my-smart-home-safe
title: "My IoT devices are no longer supported \u2013 is my smart home safe?"
total: 2
severity: medium
targets:
analyst:
name: Tier-2 analyst
role: analyst
endpoint:
category: endpoint
name: Endpoint telemetry (hb_ surfaces)
telemetry:
- endpoint
hunter:
agent: true
name: Hunt agent
network:
category: network
name: Network telemetry
telemetry:
- network
web:
category: siem
name: Web server / proxy logs
telemetry:
- network
tlp: clear
type: investigation
---
# Compromised IoT Network Activity
This hunt identifies the network-level behaviors of unmanaged IoT devices that have been compromised. It uses a gated flow, starting with a lead query that identifies hosts with high DNS failure rates, which often indicates botnet C2 or DGA activity. Only when these leads are verified does the hunt pivot into expensive behavioral analysis to find evidence of lateral movement (internal port scanning) and initial access redirection (phishing traffic). The hunt is designed for devices that lack endpoint agents by focusing on network telemetry surfaces.
## dns-failed-lookups-lead
<!-- Lead: High DNS failure volume -->
Identify internal IPs generating a high volume of NXDOMAIN or SERVFAIL responses, suggesting automated C2 or DGA attempts.
```sqlite target=endpoint role=scoping params=(lookback_days=lookback_days)
~~~yaml
expected: A list of IPs with persistent lookup failures. Silence suggests no heavy
automated C2 activity is visible on this surface.
reads:
- src_endpoint_ip
- rcode
- query_hostname
- time
silence: not_evidence_of_absence
source: hb_dns_activity
verified: dry-run
verified_at: '2026-11-21'
~~~
SELECT src_endpoint_ip, COUNT(*) as fail_count, COUNT(DISTINCT query_hostname) as unique_domains, MIN(time) as first_seen, MAX(time) as last_seen FROM hb_dns_activity WHERE rcode IN ('NXDOMAIN', 'SERVFAIL') AND time >= datetime('now', '-{{lookback_days}} days') GROUP BY src_endpoint_ip HAVING fail_count > 50 ORDER BY fail_count DESC
```
## evaluate-dns-lead
<!-- Evaluate DNS leads -->
```agent target=hunter
cite: required
context:
- dns-failed-lookups-lead
max_iterations: 3
objective: Review the DNS lookup failures and determine if the frequency and target
domains suggest an automated botnet agent.
success_criteria: A verdict for each IP citing specific failure counts.
tools:
- endpoint
- network
- web
```
## gate-on-suspicious-dns
<!-- Gate on suspicious DNS -->
if~: "the evaluate-dns-lead verdict is suspicious for at least one host" (confidence: high, judge=hunter)
then: → investigate-iot-behavior
indeterminate: → analyst-validation
unavailable: → analyst-validation (blind_spot: gateway-logging-gap)
else: → close-out-hunt
## investigate-iot-behavior
<!-- Investigate IoT behaviors -->
parallel:
- → internal-port-probing
- → http-phishing-traffic
join: → confirm-iot-compromise
## internal-port-probing
<!-- Internal management port probing -->
Detect anomalous internal scanning of management ports originating from the suspicious IPs.
```sqlite target=network role=detection-candidate params=(scope_hosts=scope_hosts, management_ports=management_ports, lookback_days=lookback_days)
~~~yaml
baseline:
compare: first_seen
window: '{{lookback_days}}d'
expected: One IP hitting multiple internal targets on ports like 23 (Telnet) or 5555
(ADB).
prevalence:
by: src_endpoint_ip
key:
- src_endpoint_ip
- dst_endpoint_port
rare_below: 3
reads:
- src_endpoint_ip
- dst_endpoint_ip
- dst_endpoint_port
- direction
- time
silence: not_evidence_of_absence
source: hb_network_connection
verified: dry-run
verified_at: '2026-11-21'
~~~
SELECT src_endpoint_ip, dst_endpoint_port, COUNT(DISTINCT dst_endpoint_ip) as internal_targets, MIN(time) as first_seen, MAX(time) as last_seen FROM hb_network_connection WHERE direction = 'outbound' AND ('{{scope_hosts}}' = '' OR instr(',' || '{{scope_hosts}}' || ',', ',' || src_endpoint_ip || ',') > 0) AND instr(',' || '{{management_ports}}' || ',', ',' || CAST(dst_endpoint_port AS TEXT) || ',') > 0 AND time >= datetime('now', '-{{lookback_days}} days') GROUP BY src_endpoint_ip, dst_endpoint_port HAVING internal_targets > 5 ORDER BY internal_targets DESC
```
## http-phishing-traffic
<!-- HTTP phishing traffic -->
Identify HTTP requests to known phishing domains, indicating a hijacked IoT gateway or router.
```sqlite target=web role=enrichment params=(scope_hosts=scope_hosts, phishing_indicators=phishing_indicators, lookback_days=lookback_days)
~~~yaml
expected: Connections to phishing infrastructure from internal IPs.
reads:
- src_endpoint_ip
- url_hostname
- user_agent
- time
silence: not_evidence_of_absence
source: hb_http_activity
verified: dry-run
verified_at: '2026-11-21'
~~~
SELECT src_endpoint_ip, url_hostname, user_agent, COUNT(*) as req_count, MIN(time) as first_seen FROM hb_http_activity WHERE ('{{scope_hosts}}' = '' OR instr(',' || '{{scope_hosts}}' || ',', ',' || src_endpoint_ip || ',') > 0) AND instr(',' || '{{phishing_indicators}}' || ',', ',' || LOWER(url_hostname) || ',') > 0 AND time >= datetime('now', '-{{lookback_days}} days') GROUP BY src_endpoint_ip, url_hostname, user_agent ORDER BY req_count DESC
```
## confirm-iot-compromise
<!-- Confirm IoT compromise -->
```agent target=hunter
cite: required
context:
- evaluate-dns-lead
- internal-port-probing
- http-phishing-traffic
max_iterations: 5
objective: Determine if the suspicious IP is a compromised IoT device by correlating
the automated DNS behavior with internal scanning and phishing redirection.
success_criteria: A verdict of malicious or suspicious for any IP showing multiple
indicators.
tools:
- endpoint
- network
- web
```
## route-to-containment
<!-- Route to containment -->
if~: "the confirm-iot-compromise verdict is malicious for at least one host" (confidence: high, judge=hunter)
then: → isolate-compromised-ip
indeterminate: → analyst-validation
unavailable: → analyst-validation (blind_spot: unmanaged-device-blind-spot)
else: → close-out-hunt
## isolate-compromised-ip
<!-- Isolate compromised IP -->
```action target=network
~~~yaml
approval: required
~~~
Isolate the identified IPs at the firewall or router. Place the devices in a quarantine VLAN to prevent lateral movement and C2 participation.
```
→ analyst-validation
## analyst-validation
<!-- Analyst validation -->
```manual target=analyst
Review the cited IPs. Use MAC OUI lookups or internal inventories to identify the device manufacturer and model. Verify if the identified devices are end-of-life.
```
→ close-out-hunt
## close-out-hunt
<!-- Hunt close-out -->
```manual target=analyst
Record all compromised devices. Recommend decommissioning for any identified end-of-life equipment. Document any network segmentation gaps found during the hunt.
```
→ end
Run this hunt against your own data.
In Huntbase, Scout runs every step across the tools you already connect, checks back through your history and returns a verdict with the evidence. The hunt.md file is open, so you can also take it anywhere that reads the format.
Drafted by machine (huntbase-hunt-generation, hb_google/gemini-3-flash-preview), checked by dry-run, lint, then reviewed by a person.