CVE-2025-59953: CWE-502: Deserialization of Untrusted Data in InternLM lmdeploy
CVE-2025-59953 is a critical remote code execution vulnerability in InternLM's lmdeploy toolkit affecting versions from 0.9.1 up to but not including 0.10.2. The vulnerability arises because the AsyncRPCServer component uses Python's pickle.loads() to deserialize incoming RPC messages without sanitization, allowing untrusted data to execute arbitrary code. This flaw is patched in version 0.10.2.
AI Analysis
Technical Summary
InternLM's lmdeploy toolkit versions >=0.9.1 and <0.10.2 implement an RPC server (AsyncRPCServer in zmq_rpc.py) that deserializes received messages using pickle.loads() without validation. This unsafe deserialization of untrusted data leads to a remote code execution vulnerability (CWE-502). The issue is fixed in version 0.10.2.
Potential Impact
An attacker can send specially crafted RPC messages to the vulnerable AsyncRPCServer, triggering arbitrary code execution on the server. This compromises confidentiality, integrity, and availability of the affected system, as reflected by the CVSS 3.1 score of 9.8 (critical).
Mitigation Recommendations
Upgrade lmdeploy to version 0.10.2 or later, where the unsafe deserialization vulnerability has been patched. No other mitigations are specified or required once the patch is applied.
CVE-2025-59953: CWE-502: Deserialization of Untrusted Data in InternLM lmdeploy
Description
CVE-2025-59953 is a critical remote code execution vulnerability in InternLM's lmdeploy toolkit affecting versions from 0.9.1 up to but not including 0.10.2. The vulnerability arises because the AsyncRPCServer component uses Python's pickle.loads() to deserialize incoming RPC messages without sanitization, allowing untrusted data to execute arbitrary code. This flaw is patched in version 0.10.2.
CVSS v3.1
Score 9.8critical
Affected software
InternLM
lmdeploy
pkg:github/internlm/lmdeployRun on your own infrastructure? Check whether these packages are installed with threat-finder — our free open-source scanner.
Weaknesses
AI-Powered Analysis
Machine-generated threat intelligence
Technical Analysis
InternLM's lmdeploy toolkit versions >=0.9.1 and <0.10.2 implement an RPC server (AsyncRPCServer in zmq_rpc.py) that deserializes received messages using pickle.loads() without validation. This unsafe deserialization of untrusted data leads to a remote code execution vulnerability (CWE-502). The issue is fixed in version 0.10.2.
Potential Impact
An attacker can send specially crafted RPC messages to the vulnerable AsyncRPCServer, triggering arbitrary code execution on the server. This compromises confidentiality, integrity, and availability of the affected system, as reflected by the CVSS 3.1 score of 9.8 (critical).
Mitigation Recommendations
Upgrade lmdeploy to version 0.10.2 or later, where the unsafe deserialization vulnerability has been patched. No other mitigations are specified or required once the patch is applied.
Technical Details
- Data Version
- 5.2
- Assigner Short Name
- GitHub_M
- Date Reserved
- 2025-09-23T14:33:49.506Z
- Cvss Version
- 3.1
- State
- PUBLISHED
Threat ID: 6aaab9fd55bf5e2cf5d03c76
Added to database: 09/16/2026, 15:47:09 UTC
Last enriched: 09/16/2026, 16:01:32 UTC
Last updated: 09/16/2026, 16:01:55 UTC
Views: 8
Community Reviews
0 reviewsCrowdsource mitigation strategies, share intel context, and vote on the most helpful responses. Sign in to add your voice and help keep defenders ahead.
Want to contribute mitigation steps or threat intel context? Sign in or create an account to join the community discussion.
Actions
Updates to AI analysis require Pro Console access. Upgrade inside Console → Billing.
Need more coverage?
Upgrade to Pro Console for AI refresh and higher limits.
For incident response and remediation, OffSeq services can help resolve threats faster.
Latest Threats
Check if your credentials are on the dark web
Instant breach scanning across billions of leaked records. Free tier available.