Executive brief
LightGBM is a machine learning library used to build prediction models for classification and ranking tasks. An attacker can craft a malicious model file that causes the library to write data outside allocated memory during prediction, potentially leading to application crashes or arbitrary code execution when processing untrusted model files.
Technical details
LightGBM through version 4.7.0 fails to validate child and split array values when parsing text-format model files. During SHAP (feature contribution) prediction calculations, invalid node references in a crafted model can trigger out-of-bounds memory writes to the leaf_depth_ buffer at attacker-controlled offsets. The vulnerability requires loading a malicious model file and performing prediction operations; no authentication or network access is required if the application loads untrusted models locally. An attacker can achieve memory corruption, denial of service, or potentially code execution depending on the memory layout and application context.
Affected products
- LightGBM LightGBM through 4.7.0
Timeline
- 2026-09-16: disclosed