Unauthenticated code execution via pickle deserialization
Published Aug 19, 2026 · Updated Aug 20, 2026
Deserialization of untrusted data in InternLM LMDeploy before 0.16.0 allows remote attackers to execute code via crafted peer messages. In disaggregated serving, handle_zmq_recv in lmdeploy/pytorch/disagg/conn/engine_conn.py calls recv_pyobj(), which runs pickle.loads() on bytes from a peer-selected ZMQ endpoint before any DistServeCacheFreeRequest validation occurs. The reachable path is an unauthenticated POST to /distserve/p2p_connect, which makes the engine connect its PULL socket to a caller-chosen address; deployments without disaggregated serving enabled are not affected.
Summary
What happened
Deserialization of untrusted data in InternLM LMDeploy before 0.16.0 allows remote attackers to execute code via crafted peer messages. In disaggregated serving, handle_zmq_recv in lmdeploy/pytorch/disagg/conn/engine_conn.py calls recv_pyobj(), which runs pickle.loads() on bytes from a peer-selected ZMQ endpoint before any DistServeCacheFreeRequest validation occurs. The reachable path is an unauthenticated POST to /distserve/p2p_connect, which makes the engine connect its PULL socket to a caller-chosen address; deployments without disaggregated serving enabled are not affected.
The record
- CVE
- CVE-2026-76850
- Published
- Aug 19, 2026
- Updated
- Aug 20, 2026
- Vendor
- InternLM
- Product
- LMDeploy
- Classifications
- CWE-502, T1190
- Attack vector
- network
- Privileges
- unauthenticated
Timeline
How it unfolded
- Aug 19, 2026CVE publishedPublication date reported by the CVE source.
- Aug 20, 2026Record updatedLatest update available in the CVE record.
Exploitability
Present is not the same as exploitable
Compare your product and version with the public record. A matching version still requires validation against your environment.
Is a vulnerable build present?
Compare these published version ranges with your installed build and any vendor patches.
- Affected versionversion=0.9.2 <0.16.0
What conditions does exploitation require?
What is affected?
Attacks
What attackers are doing with it
Daily unique IPs observed by Shadowserver honeypots for known exploited vulnerabilities (KEVs). Missing observations do not establish an absence of attacks.
Weakness, pattern, technique
Public exploit references
- [Bug] Unauthenticated pickle deserialization via ZMQ in disaggregated serving → RCEproof of concept · demonstrated
Labels summarize the accepted research assessment. They do not indicate a test against your environment.
Technologies
Your stack
See the directory against your own environment.
Your stack
Check the software in your environment
Book a demo to see how Hinoki identifies affected software and validates exploitability in your environment.
Book a demo