Live › vulnerability › CVE-2026-112075
Go to dashboardUnsloth up to 0.1.501-beta uses attacker-influenced input to choose classes to load in FastSentenceTransformer._is_transformer_module_ref, which can lead to code being run. The sources describe the flaw as remotely triggerable and say a public exploit exists.
Summary written from the sources below. Check the originals before acting.
First seen by InfraStack 2026-10-23 21:57 UTC. All times on this page are UTC.
Teams that run unsloth up to 0.1.501-beta and load sentence transformer models through FastSentenceTransformer, especially where the model name, config or module reference comes from users or untrusted sources. Teams on the patched version, or who only load fixed, locally vetted models, are less exposed; the sources do not say which input path is needed.
No product ranges from the sources yet.
pip show unsloth and check whether the version is 0.1.501-beta or lowerFastSentenceTransformer from unsloth/models/sentence_transformer.py on model or config input from outside the team0.1.902-beta or later (patch de801746386a53a0cb26216e715a0af863b76ad9)Each row is the first time that source carried this signal. The analyst’s rows are its own findings.
A weakness has been identified in unslothai unsloth up to 0.1.501-beta. This affects the function FastSentenceTransformer._is_transformer_module_ref of the file unsloth/models/sentence_transformer.py. This manipulation of the argument Type causes use of externally-controlled input to select classes or code. The attack may be initiated remotely. The exploit has been made available to the public and could be used for attacks. Upgrading to version 0.1.902-beta is able to mitigate this issue. Patch name: de801746386a53a0cb26216e715a0af863b76ad9. The affected component should be upgraded.
This page as .md · .json · in your agent: get_signal("CVE-2026-112075")
NVD data is provided by NIST and is public domain. Posts quoted belong to their authors.