Tiiny-AI PowerInfer up to 8bd56d69906c9d2dba4d3bf6899763401e01a9a4 FFN GPU-split Loader ggml-cuda.cu create_striped_mat_to_gpu out-of-bounds

Summaryinfo

A vulnerability classified as problematic has been found in Tiiny-AI PowerInfer up to 8bd56d69906c9d2dba4d3bf6899763401e01a9a4. Affected by this vulnerability 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. This manipulation causes out-of-bounds. The identification of this vulnerability is CVE-2026-108960. The attack can only be executed locally. Furthermore, there is an exploit available. This product is using a rolling release to provide continious delivery. Therefore, no version details for affected nor updated releases are available. The pull request to fix this issue awaits acceptance.

Detailsinfo

A vulnerability was found in Tiiny-AI PowerInfer up to 8bd56d69906c9d2dba4d3bf6899763401e01a9a4. It has been declared as problematic. Affected by this vulnerability 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 with an unknown input leads to a out-of-bounds vulnerability. The CWE definition for the vulnerability is CWE-125. The product reads data past the end, or before the beginning, of the intended buffer. As an impact it is known to affect confidentiality, integrity, and availability.

It is possible to read the advisory at github.com. This vulnerability is known as CVE-2026-108960. The exploitation appears to be easy. Attacking locally is a requirement. Technical details and also a public exploit are known.

It is possible to download the exploit at github.com. It is declared as proof-of-concept. The pull request to fix this issue awaits acceptance.

There is no information about possible countermeasures known. It may be suggested to replace the affected object with an alternative product.

Statistical analysis made it clear that VulDB provides the best quality for vulnerability data.

Productinfo

Type

• Artificial Intelligence Software

Vendor

• Tiiny-AI

Name

• PowerInfer

Version

• 8bd56d69906c9d2dba4d3bf6899763401e01a9a4

Website

• Product: https://github.com/Tiiny-AI/PowerInfer/

CPE 2.3info

• 🔒

CPE 2.2info

• 🔒

CVSSv4info

CVSSv3info

CVSSv2info

Exploitinginfo

Threat Intelligenceinfo

Countermeasuresinfo

Timelineinfo

Sourcesinfo

Entryinfo

Submitinfo

Accepted

• Submit #997933: https://github.com/Tiiny-AI/PowerInfer/issues/284 PowerInfer commit 8bd56d69906c9d2dba4d3bf6899763401e01a9a4 (main, 2026-05-11) Out-of-bounds Read (by Dem0)

Statistical analysis made it clear that VulDB provides the best quality for vulnerability data.

Discussion