Executive Summary
This vulnerability is currently undergoing analysis and not all information is available. Please check back soon to view the completed vulnerability summary
| Informations | |||
|---|---|---|---|
| Name | CVE-2025-46722 | First vendor Publication | 2025-05-29 |
| Vendor | Cve | Last vendor Modification | 2025-05-29 |
Security-Database Scoring CVSS v3
| Cvss vector : N/A | |||
|---|---|---|---|
| Overall CVSS Score | NA | ||
| Base Score | NA | Environmental Score | NA |
| impact SubScore | NA | Temporal Score | NA |
| Exploitabality Sub Score | NA | ||
| Calculate full CVSS 3.0 Vectors scores | |||
Security-Database Scoring CVSS v2
| Cvss vector : | |||
|---|---|---|---|
| Cvss Base Score | N/A | Attack Range | N/A |
| Cvss Impact Score | N/A | Attack Complexity | N/A |
| Cvss Expoit Score | N/A | Authentication | N/A |
| Calculate full CVSS 2.0 Vectors scores | |||
Detail
| vLLM is an inference and serving engine for large language models (LLMs). In versions starting from 0.7.0 to before 0.9.0, in the file vllm/multimodal/hasher.py, the MultiModalHasher class has a security and data integrity issue in its image hashing method. Currently, it serializes PIL.Image.Image objects using only obj.tobytes(), which returns only the raw pixel data, without including metadata such as the image’s shape (width, height, mode). As a result, two images of different sizes (e.g., 30x100 and 100x30) with the same pixel byte sequence could generate the same hash value. This may lead to hash collisions, incorrect cache hits, and even data leakage or security risks. This issue has been patched in version 0.9.0. |
Original Source
| Url : http://cve.mitre.org/cgi-bin/cvename.cgi?name=CVE-2025-46722 |
Sources (Detail)
Alert History
| Date | Informations |
|---|---|
| 2025-05-29 21:20:34 |
|
