/

CVE-2026-33356

Severity:

HIGH RISK

(7.7)

Improper Authorization in Meari IoT Cloud MQTT Broker

CVE-2026-33356

Deferred

Vulnerability type

Category:

Improper Authorization

CWE:

CWE-639

Impact:

Cross-Tenant Data Exposure in IoT Telemetry

Severity:

HIGH RISK

(7.7)

CVSS version:

3.1

Catch this CVE while code is being written

Detect vulnerable code directly in pull requests, so CVEs are fixed before merge.

Summary

Meari IoT Cloud MQTT Broker deployments running EMQX 4.x allow any authenticated low-privilege account to subscribe to global wildcard topics and receive telemetry from devices it does not own.

While publish operations are scoped correctly, insufficient authorization checks on subscription topics let users passively access other tenants' device data.

Affected context

This vulnerability appears in Meari IoT Cloud deployments that use EMQX 4.x as the MQTT broker with multi-tenant, per-device access controls.

The broker correctly restricts which topics a user can publish to, but does not apply equivalent per-device authorization to subscription requests, particularly for wildcard topics (e.g., multi-device or global streams).

Environments where tenants share a common broker namespace and rely on topic naming conventions for isolation are most affected, especially when telemetry topics use broad or predictable naming patterns.

Exploit preconditions

  • Attack surface: MQTT broker exposed to the internet or shared network for IoT devices and user applications.
  • Required access: Any valid, low-privilege Meari IoT Cloud account with MQTT credentials.
  • Trigger point: Attacker issues SUBSCRIBE operations to global or wildcard topics that aggregate telemetry across many devices.
  • Why exploitation is possible: Authorization logic enforces publish restrictions but does not symmetrically validate subscriptions against a per-user/per-device access control list, allowing overbroad read access.

Why this happens

The core issue is asymmetric authorization between publish and subscribe paths in the MQTT access control layer.

Topic-authorization policies appear to be applied only to write operations, with subscription requests either not checked or checked using a more permissive rule set.

This maps to CWE-639: Insecure Direct Object Reference / Improper Authorization, where user-controlled references (here, MQTT topics and wildcards) are trusted without verifying that the authenticated principal is allowed to access the underlying resource (another tenant's device telemetry).

The insecure pattern is relying on topic naming and authentication alone, without enforcing per-tenant or per-device policy for every subscribe operation, including wildcard forms.

Potential impact

  • Unauthorized access to live telemetry from other users' IoT devices.
  • Cross-tenant privacy violations, potentially exposing household activity patterns, sensor readings, or camera-associated metadata.
  • Collection and profiling of large-scale environmental or behavioral data by a single malicious tenant.
  • Intelligence gathering for follow-on attacks, such as identifying device types, firmware versions, or operational schedules.
  • Regulatory and contractual non-compliance due to unintended data sharing between tenants.

General mitigation guidance

Mitigation generally involves enforcing symmetrical, least-privilege authorization for both publish and subscribe operations across all MQTT topics:

  • Apply per-tenant and per-device access control lists (ACLs) to every SUBSCRIBE request, including wildcard topics, ensuring a client can only subscribe to topics for resources it owns or is explicitly granted.
  • Use structured topic namespaces that encode tenant and device identifiers in a way that can be reliably mapped to identity and authorization policies.
  • Prefer deny-by-default policies for subscriptions, only granting access to explicitly allowed topics rather than broad patterns.
  • Where possible, segregate tenants logically or physically (e.g., per-tenant virtual hosts, namespaces, or separate brokers) to reduce cross-tenant exposure if topic policies fail.
  • Implement and enforce privacy-by-design defaults, limiting exposure of sensitive telemetry fields and avoiding unnecessary broadcast channels.

How teams detect issues like this

Effective detection focuses on both configuration and behavioral analysis:

  • Configuration and policy review: Examine MQTT ACLs and broker configuration for differences between publish and subscribe rules, especially any global or wildcard read permissions.
  • Static / policy analysis: Use automated checks or scripts to validate that all topic patterns, including wildcards, are governed by explicit authorization rules that tie topics to tenant identities.
  • Log analysis and monitoring: Monitor subscription logs for clients subscribing to unusually broad topic filters (e.g., multi-level wildcards or cross-tenant paths) and correlate with their expected device ownership.
  • Access pattern anomaly detection: Flag accounts that receive telemetry from large numbers of devices or from devices not normally associated with their tenant.
  • Security testing: Include MQTT authorization testing in routine assessments, validating that a low-privilege client cannot subscribe to other tenants' topics using wildcards or guessed topic names.

Sources

SourceLink
github.comgithub.com/xn0tsa/nobody-puts-baby-in-a-corner
runzero.comrunzero.com/advisories/meari-mqtt-broker-missing-per-device-subscribe-acl-cve-2026-33356/

Frequently asked

questions

Yes. Any attacker who can obtain valid, low-privilege Meari IoT Cloud MQTT credentials and connect to the broker can exploit it remotely by subscribing to overbroad topics.

The attacker needs only an authenticated account with standard MQTT connection permissions; elevated or administrative roles are not required. The weakness lies in over-permissive subscription authorization.

Implement strict, symmetric authorization checks for every SUBSCRIBE and PUBLISH operation, ensuring that topic filters (including wildcards) are evaluated against a per-user or per-tenant access control model that maps topics to owned devices.

Design MQTT authorization with least privilege from the start, avoid relying solely on topic naming conventions, regularly test cross-tenant isolation, and enforce security policies via automated checks on broker configuration and access control rules before deployment.

Find out what's already
exploitable in your codebase.