Tools AI Risk Radar ai-incident-0250
Incident record
Forged vLLM shared-memory handles could execute code
- Severity
- High
- Status
- Patched
- Type
- Infra Vuln
- Target
- vllm-project/vllm
- Actor
- researcher
What happened
An authenticated caller could forge a multimodal cache handle and reach unsafe deserialisation in a worker process.
The path requires shared-memory multimodal caching. The researcher describes a reproduced chain through public serving routes, with insecure serialisation disabled. The maintainer rates the issue High, CVSS 8.8, and lists versions through 0.25.1 as affected and 0.32.0 as patched.
Sources
- Maintainer security advisorygithub.com/vllm-project/vllm/security/advisories/GHSA-6757-h…
One record from the AI Risk Radar, maintained by the Security Desk.Data: CSV · JSON ·RSS · CC BY 4.0 with attribution to YFarmX.Tracker updated · 28 October 2026