Live Exposure Defense: From CVE to Confirmed Exposure in 12 Hours – See more

Back
New CVE Detected

CVE-2025-59953 – Unauthenticated RCE via Pickle Deserialization – LMDeploy 0.9.1 to 0.10.1

Be the first to know when new zero-days emerge:

Subscribe to Email Alerts
Subscribe to RSS Feed

Summary

CVE-2025-59953 is a critical (CVSS 9.8) remote code execution vulnerability in LMDeploy, an open-source toolkit for compressing, deploying, and serving large language models. The flaw resides in LMDeploy’s AsyncRPCServer (lmdeploy/pytorch/engine/mp_engine/zmq_rpc.py), which deserializes incoming network messages with Python’s pickle.loads() without any validation. An unauthenticated, network-adjacent attacker can exploit this to execute arbitrary code on the host running the RPC server.

Technical details

  • Root cause: The call_and_response() function in the RPC server implementation calls pickle.loads() directly on data received over the network, with no sanitization, type checking, or allow-listing. Python’s pickle format allows arbitrary code execution during object deserialization by design, making this pattern inherently unsafe when applied to untrusted input.
  • Trigger conditions: LMDeploy’s AsyncRPCServer is used to coordinate inter-process/multi-node tensor-parallel and data-parallel inference orchestration via ZeroMQ (ZMQ). Prior to the fix, the RPC server bound to tcp://*, exposing it on all network interfaces rather than restricting it to localhost.
  • Attack vector: Network (AV:N), no authentication or user interaction required. An attacker who can reach the dynamically assigned RPC port — for example, by scanning a network segment where LMDeploy is deployed — can send a crafted malicious pickle payload directly to the endpoint.
  • Impact: Successful exploitation results in arbitrary remote code execution with the privileges of the LMDeploy process. This can lead to full host compromise, lateral movement, exfiltration of proprietary model weights or sensitive data, and disruption of the inference service.

Affected software

  • Product: InternLM LMDeploy
  • Affected versions: 0.9.1 through 0.10.1 (inclusive)
  • Patched version: 0.10.2 and later

Severity

  • CVSS v3.1 Base Score: 9.8 (Critical)
  • Vector: CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:H/A:H

Mitigation and recommended actions

  • Immediate: Upgrade LMDeploy to version 0.10.2 or later, which changes the RPC server binding from tcp://* to localhost, preventing remote network access to the RPC endpoint.
  • If immediate patching is not possible:
    • Do not expose the LMDeploy RPC port to untrusted networks; restrict access using firewall rules or network segmentation so only trusted local processes/hosts can reach it.
    • Run the LMDeploy service under a restricted, low-privilege user account to limit the impact of a potential compromise.
    • Use container network namespaces without host networking to isolate the RPC server from the broader network.
    • Monitor for unexpected inbound connections or unusual process activity on hosts running LMDeploy inference services.

References

Are you exposed?

Get a free report of your organization’s exposure to this CVE and threat

  • Mapping of all Assets with this Technology
  • Identification of potentially exposed assets to this CVE
  • Confirmation of verified exploitable assets
Get Exposure Report

How IONIX’s External Exposure Management Platform Detects and Validates
Zero-Days to Shrink MTTR

1

Map your entire attack surface (continously)

IONIX uses multi-factor discovery methods, including DNS analysis, certificate mapping, metadata inspection, and more, to automatically map every internet-facing asset across your environment. This includes cloud instances, third-party platforms, shadow IT, and even forgotten infrastructure that traditional tools miss.

2

Monitor for new CVEs

Dozens of threat intel feeds using agentic technology are continuously analyzed to detect the appearance of proof-of-concept code, exploit kits, and indicators of active targeting. IONIX goes further by applying AI to proactively evaluate whether emerging vulnerabilities are likely to be exploited, even before PoCs go public.

3

Identify Potential External Exposures

Not all CVEs matter. IONIX filters vulnerabilities by asking attacker-centric questions: Can it be reached from the internet? Does it require authentication? Is it being exploited in the wild? This dramatically reduces noise and focuses teams on threats that can actually be weaponized.

4

Create Safe, Scalable Exploit Validations

IONIX transforms real-world PoCs into safe, non-intrusive test payloads that can be run in production environments without disruption. These simulations are precisely targeted to the systems that are vulnerable, ensuring rapid validation without unnecessary load.

5

Execute Exploit Validations

By combining context about software stack, versioning, exposure status, and reachability, IONIX ensures that only the right payloads are executed against the right assets, maximizing efficiency and minimizing risk.

6

Drive Fast and Actionable Remediation

Results are routed through integrations with ticketing, SOAR, and SIEM tools. Issues are written in plain language, bundled into remediation clusters, and prioritized based on asset criticality, exploitability, and blast radius. This shortens mean time to remediation (MTTR) and empowers teams to act with confidence.

Are you exposed?

Get a free report of your organization’s exposure to this CVE and threat

  • Mapping of all Assets with this Technology
  • Identification of potentially exposed assets to this CVE
  • Confirmation of verified exploitable assets
Get Exposure Report
Close

Subscribe to Threat Center RSS

Copy/paste the link below into your preferred RSS reader or follow these instructions to subscribe to Slack alerts.

Open RSS feedSlack setup instructions

Close

Get Real-Time CVE Alerts to Your Email

Be the first to know when new zero-days emerge