CVE-2026-108960
5.3MediumPublished 13 Oct 2026 · NVD status: Deferred · AI triage by gemini-3.8-flash
- Attack vector
- Local
- Complexity
- Low
- Privileges needed
- Low
- User interaction
- None
- Impact
- Limited impact
- Weakness
- CWE-119, CWE-125
Impact
A local attacker can trigger an out-of-bounds read to crash the process or expose local memory. They need local shell or system access on the machine running PowerInfer.
Root cause
The create_striped_mat_to_gpu function in ggml-cuda.cu reads beyond assigned memory boundaries during GPU matrix loading.
Affected systems
Tiiny-AI PowerInfer up to commit 8bd56d69906c9d2dba4d3bf6899763401e01a9a4.
CVSS v3.1 · CVSS:3.1/AV:L/AC:L/PR:L/UI:N/S:U/C:L/I:L/A:L
Fix & remediation
No official patched release or merged commit is available yet; check the repository and apply the pending pull request or wait for its merge. As a temporary mitigation, restrict local user access to the host running the GPU loader.
$ git -C /path/to/PowerInfer rev-parse HEADAI-suggested. Run it in staging first and check it against the vendor advisory. Why?
NVD description
A vulnerability was detected in Tiiny-AI PowerInfer up to 8bd56d69906c9d2dba4d3bf6899763401e01a9a4. Affected is the function create_striped_mat_to_gpu of the file smallthinker/ggml/src/ggml-cuda/ggml-cuda.cu of the component FFN GPU-split Loader. The manipulation results in out-of-bounds read. The attack needs to be approached locally. The exploit is now public and may be used. This product utilizes a rolling release system for continuous delivery, and as such, version information for affected or updated releases is not disclosed. The pull request to fix this issue awaits acceptance.