CVE-2026-103395 LightLLM through 1.2.0 Unauthenticated Remote Code Execution via Visual-Only RPyC Service
Attackers can execute arbitrary code on LightLLM visual-only deployments by sending malicious serialized objects to an unauthenticated RPyC service.
- Exploited in the wild
- Not known
- Login needed
- Not needed
- Attackable
- Over the internet
- Victim action needed
- Not needed
- Fix
- None listed
What is CVE-2026-103395?
CVE-2026-103395 is a critical unsafe deserialization vulnerability in ModelTC LightLLM, published on 30 Sep 2026. An attacker can exploit it over the internet without logging in, with no action from the victim.
Official description
LightLLM through 1.2.0 visual_only deployments expose an unauthenticated RPyC service with allow_pickle enabled that deserializes attacker-supplied arguments in the remote_infer_images method. Attackers can reach the visual RPyC port and pass objects with __reduce__ methods to execute arbitrary code with service account privileges.
Weakness: CWE-502
Which versions are affected by CVE-2026-103395?
How do I fix CVE-2026-103395?
No fix has been published for CVE-2026-103395 yet. Limit exposure of the affected software until one is available.
Upgrade, and only deserialise data from trusted sources or with an allow-list of types. More about Unsafe deserialization.
Is CVE-2026-103395 being exploited?
It is not confirmed as exploited in the wild, but public exploit code for CVE-2026-103395 exists.
Sources
Also known as GHSA-mw2v-h3wh-mp72.