Database/AI/ML frameworks & serving

llama.cpp: integer overflow in GGUF file parsing lets a malicious model file run code
Impact
A crafted GGUF model file causes an integer overflow before a memory write in llama.cpp's GGUF parser. ZDI says this can be used to run arbitrary code in the context of the process that loads the file. On a serving node, that process usually holds the GPU, the model cache and any credentials the server was started with. Anywhere users or pipelines can supply their own GGUF files, such as a shared inference service, a model-pull workflow or a tenant notebook, loading an untrusted model is enough. ZDI fixed this in the same pull request as CVE-2026-126975, but it is a separate flaw (integer overflow, not a missing length check).
Who can reach it
Anyone who can get a crafted GGUF file loaded by llama.cpp, for example by uploading or pointing the server at a model, or by getting an operator to open one. The scoring requires user interaction (loading the file) but no authentication to llama.cpp itself.
What to do
Rebuild or update llama.cpp to a build that includes ggml-org/llama.cpp pull request 31336, then restart every server or job that embeds it. No reboot is needed. The record does not name a fixed release tag. Until then, load GGUF files only from trusted sources.
References
Related entries
- llama.cpp: heap buffer overflow in GGUF file parsing lets a malicious model file run codeCVE-2026-126975 · llama.cpp (GGUF model file parser)High
- timm: untrusted checkpoint deserialization executes arbitrary code in the loading processCVE-2026-15679 · Hugging Face pytorch-image-models (timm) checkpoint loadingHigh
- HuggingFace transformers (`Trainer._load_rng_state`): Arbitrary code execution when a training run resumesCVE-2026-1839 · HuggingFace transformers (`Trainer._load_rng_state`)High
- NVIDIA NeMo Speech: malicious input data leads to remote code executionCVE-2026-24239 · NVIDIA NeMo SpeechHigh
- NVIDIA NeMo Speech: RCE and privilege escalation in the speech data explorerCVE-2026-24267 · NVIDIA NeMo Speech (speech data explorer)High
- BentoML (`safe_extract_tarfile`): Tar extraction escape despite the "safe" helperCVE-2026-27905 · BentoML (`safe_extract_tarfile`)High
This entry is curated: imported from vendor advisories with machine assistance, not yet individually verified. Confirm against your vendor's advisory before acting, and report anything wrong.