CVE-2025-66455

CRITICAL CVSS 9.8 PoC Available
Published: 2026-09-18
Modified: 2026-09-18
Enriched via mock
Tenable Plugin

Checking the public Tenable catalog…

AI Threat Summary

CVE-2025-66455 is a critical vulnerability with a CVSS score of 9.8 affecting the target software. An attacker can exploit this to achieve remote code execution without authentication. Threat hunters should monitor for exploit attempts and apply vendor patches immediately.

Description

LMDeploy is a toolkit for compressing, deploying, and serving large language models. Starting in version 0.9.2 and prior to version 0.16.0, LMDeploy's PyTorch DistServe/PD-disaggregation control plane used `recv_pyobj()` to deserialize messages received through a ZeroMQ PULL socket. PyZMQ implements `recv_pyobj()` using Python pickle deserialization, which can execute arbitrary code while reconstructing an object. The peer address used by the receiver was supplied through the `POST /distserve/p2p_connect` HTTP endpoint. An attacker who could reach an affected DistServe API server could cause the server to connect to an attacker-controlled ZeroMQ endpoint and deserialize a crafted pickle payload. API-key authentication is not enabled unless the operator explicitly configures it. As a result, affected DistServe deployments without API keys allowed unauthenticated remote code execution with the privileges of the LMDeploy serving process. This issue affects the PyTorch backend when PD-disaggregation/DistServe is enabled. Ordinary deployments that do not use the affected disaggregated-serving path do not expose this data flow. The fix was released in LMDeploy 0.16.0. Users who cannot upgrade immediately should prevent untrusted clients from reaching `/distserve/*` endpoints, restrict the DistServe HTTP and ZeroMQ control planes to trusted cluster networks, configure API-key authentication, and block arbitrary outbound ZeroMQ connections from serving nodes. These measures reduce exposure but do not make pickle deserialization safe.

Attack Chain
1
Reconnaissance

Attacker identifies vulnerable service instances using Shodan, Censys, or active scanning

2
Initial Access

Attacker crafts malicious payload exploiting CVE-2025-66455 to achieve remote code execution

3
Execution

Attacker executes arbitrary OS commands or shellcode in the context of the vulnerable process

4
Persistence

Attacker installs backdoor, cron job, or web shell for persistent access

5
Lateral Movement

Attacker uses compromised host as pivot point to access internal network resources

MITRE ATT&CK Techniques
T1190 Exploit Public-Facing Application
Initial Access

Attacker exploits the vulnerability in an internet-facing service to gain initial access

T1059 Command and Scripting Interpreter
Execution

Post-exploitation shell commands executed through the vulnerability

T1105 Ingress Tool Transfer
Command and Control

Attacker downloads malware or tools after gaining code execution

Indicators of Compromise
Type Value Description
user_agent Mozilla/5.0 (compatible; exploit/1.0) Generic exploit scanner user agent pattern
url /api/vuln-endpoint?payload=<malicious> Exploit attempt against application endpoint — monitor access logs for anomalous params
registry_key HKLM\SOFTWARE\Microsoft\Windows\CurrentVersion\Run\backdoor Persistence mechanism after successful RCE
Remediation
1. Apply vendor-supplied security patches immediately. 2. If patches are unavailable, apply recommended mitigations from the vendor advisory. 3. Enable WAF rules or IDS signatures to detect exploitation attempts. 4. Monitor logs for indicators of compromise listed above. 5. Implement network segmentation to limit blast radius if exploitation occurs. 6. Subscribe to vendor security advisories for this product.
Quick Facts
CVE ID CVE-2025-66455
CVSS Score 9.8
Vector CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:H/A:H
Published 2026-09-18
Exploited No
CISA KEV No
PoC Available