Skip to main content

Azure Connected Machine Agent CVE-2025-47989

HIGH
Improper Access Control (CWE-284)
2025-10-14 secure@microsoft.com
7.0
CVSS 3.1 · Vendor: microsoft
Share

Severity by source

Vendor (microsoft) PRIMARY
7.0 HIGH
AV:L/AC:H/PR:L/UI:N/S:U/C:H/I:H/A:H
vuln.today AI
7.0 HIGH

Local EoP needing an existing low-privileged account (PR:L) and difficult-to-meet timing conditions (AC:H), with full compromise of the host on success (C/I/A:H).

3.1 AV:L/AC:H/PR:L/UI:N/S:U/C:H/I:H/A:H
4.0 AV:L/AC:H/AT:N/PR:L/UI:N/VC:H/VI:H/VA:H/SC:N/SI:N/SA:N

Primary rating from Vendor (microsoft).

CVSS VectorVendor: microsoft

Attack Vector
Local
Attack Complexity
High
Privileges Required
Low
User Interaction
None
Scope
Unchanged
Confidentiality
High
Integrity
High
Availability
High

Lifecycle Timeline

2
Analysis Generated
Oct 08, 2026 - 13:01 vuln.today
CVE Published
Oct 14, 2025 - 17:15 cve.org
HIGH 7.0

DescriptionCVE.org

Improper access control in Azure Connected Machine Agent allows an authorized attacker to elevate privileges locally.

AnalysisAI

Local privilege escalation in the Azure Connected Machine Agent (azcmagent, the Azure Arc onboarding agent) allows an authorized attacker to gain elevated privileges on an affected host. Exploitation requires the attacker to already hold an authenticated, low-privileged local account (PR:L) on a machine where the agent is installed and running; the vector is strictly local (AV:L) with high attack complexity (AC:H) and no user interaction (UI:N), and it can deliver high confidentiality, integrity, and availability impact, yielding a CVSS 3.1 base score of 7.0. EPSS is low at 0.56% (45th percentile), and no public exploit code or confirmed active exploitation was identified at time of analysis, so this is a moderate-priority local escalation rather than a top-tier emergency.

Technical ContextAI

This is a CWE-284 (Improper Access Control) defect in Microsoft's Azure Connected Machine Agent, the software installed on servers and workstations to onboard them to Azure Arc. Because the agent typically runs as a privileged service and manages configuration and credential material, weak access controls over agent-owned resources (files, registry keys, named pipes, or service interfaces) can let a low-privileged local user act with the agent's higher privileges. The high attack complexity (AC:H) in the assessed CVSS 3.1 vector implies the weakness is not a trivial, deterministic privilege grant - it likely depends on a race condition, precise timing, or a specific non-default state the attacker cannot reliably control, which reduces the chance of dependable, repeatable exploitation. The CVSS vector (AV:L/AC:H/PR:L/UI:N/S:U/C:H/I:H/A:H) confirms a local, post-foothold escalation requiring an authenticated low-privileged session with no UI interaction; the CPE entry cpe:2.3:a:microsoft:azure_connected_machine_agent with wildcard versioning indicates the flaw is tracked against the agent as a whole rather than one fixed release. The specific configuration or feature that exposes the improper access control is not detailed in the provided data and should be confirmed against the MSRC advisory.

RemediationAI

Apply the vendor fix for the Azure Connected Machine Agent by upgrading to the patched release documented in the Microsoft MSRC update guide at https://msrc.microsoft.com/update-guide/vulnerability/CVE-2025-47989; the provided data confirms an advisory and patch path but does not include an exact fix version, so treat this as patch-available-per-vendor-advisory and confirm the specific version string on the MSRC page before rollout. Because exploitation is local and requires an existing low-privileged account (PR:L) on a machine running the agent, the most effective compensating controls are reducing the local attack surface on Arc-onboarded hosts: audit and remove unnecessary local interactive accounts and restrict who may log on locally to machines running azcmagent, since the attack cannot be launched remotely and an initial local foothold is a hard prerequisite. Where immediate patching is not possible, prioritize the most exposed hosts, monitor for unexpected privilege changes or suspicious child processes spawned by the agent service, and tighten permissions on agent-owned files, services, and communication channels to make the access-control gap harder to reach. Note that restricting local logon may impede legitimate administrative workflows, and monitoring or ACL tightening are detection- and hardening-only measures that do not remove the underlying CWE-284 weakness - patching remains the only complete fix.

Share

CVE-2025-47989 vulnerability details – vuln.today

This site uses cookies essential for authentication and security. No tracking or analytics cookies are used. Privacy Policy