LinuxvLLM CUDA Kernel Integer Overflow Leaks Cross-Tenant Inference Prompts
An integer overflow flaw in vLLM PagedAttention CUDA kernels allows concurrent inference requests to read sensitive prompt tokens across multi-tenant GPU memory.
Topic
2 articles
CyberSecureToday covers CWE-190 across 2 articles published between 25 Sep 2026 and 27 Sep 2026.
The coverage includes 2 vulnerability advisories. 2 advisories are rated critical or high severity.
Stories on CWE-190 most often intersect with Artificial Intelligence, vLLM and CUDA, with Technology, Cloud & SaaS the most affected sector. The most recent story, “vLLM CUDA Kernel Integer Overflow Leaks Cross-Tenant Inference Prompts”, was published 27 Sep 2026. Every article is listed below, newest first, with the affected products, severity and remediation guidance in each piece.
LinuxAn integer overflow flaw in vLLM PagedAttention CUDA kernels allows concurrent inference requests to read sensitive prompt tokens across multi-tenant GPU memory.
Threat IntelA critical integer overflow flaw CVE-2026-86289 in the Ollama GGUF model decoder allows attackers to trigger memory corruption and remote code execution on AI inference clusters.