NVIDIA Triton OOB Read via Python Backend
CVE-2025-23333 Published on August 6, 2025
NVIDIA Triton Inference Server for Windows and Linux contains a vulnerability in the Python backend, where an attacker could cause an out-of-bounds read by manipulating shared memory data. A successful exploit of this vulnerability might lead to information disclosure.
Vulnerability Analysis
CVE-2025-23333 can be exploited with network access, and does not require authorization privileges or user interaction. This vulnerability is consided to have a high level of attack complexity. The potential impact of an exploit of this vulnerability is considered to have a high impact on confidentiality, with no impact on integrity and availability.
Weakness Type
Out-of-bounds Read
The software reads data past the end, or before the beginning, of the intended buffer. Typically, this can allow attackers to read sensitive information from other memory locations or cause a crash. A crash can occur when the code reads a variable amount of data and assumes that a sentinel exists to stop the read operation, such as a NUL in a string. The expected sentinel might not be located in the out-of-bounds memory, causing excessive data to be read, leading to a segmentation fault or a buffer overflow. The software may modify an index or perform pointer arithmetic that references a memory location that is outside of the boundaries of the buffer. A subsequent read operation then produces undefined or unexpected results.
Products Associated with CVE-2025-23333
You can be notified by email with stack.watch whenever vulnerabilities like CVE-2025-23333 are published in NVIDIA Triton Inference Server:
Affected Versions
NVIDIA Triton Inference Server Version All versions prior to 25.07 is affected by CVE-2025-23333Exploit Probability
EPSS (Exploit Prediction Scoring System) scores estimate the probability that a vulnerability will be exploited in the wild within the next 30 days. The percentile shows you how this score compares to all other vulnerabilities.