CVE-2026-93989
Severity CVSS v4.0:
LOW
Type:
Unavailable / Other
Publication date:
19/09/2026
Last modified:
28/09/2026
Description
vLLM through 0.29.0 fails to properly validate bad_words token indices against the model's generation output width in SamplingParams.update_from_tokenizer(). Attackers can supply out-of-bounds token indices that corrupt logits memory of concurrent requests, causing different in-flight HTTP requests to return incorrect tokens.
Impact
Base Score 4.0
2.30
Severity 4.0
LOW
Base Score 3.x
3.10
Severity 3.x
LOW
Vulnerable products and versions
| CPE | From | Up to |
|---|---|---|
| cpe:2.3:a:vllm:vllm:*:*:*:*:*:*:*:* | 0.29.0 (including) |
To consult the complete list of CPE names with products and versions, see this page
References to Advisories, Solutions, and Tools
- https://github.com/vllm-project/vllm
- https://github.com/vllm-project/vllm/blob/98dff2a81d747d1dba01a47f939f48c3526d4206/vllm/sampling_params.py#L694-L753
- https://github.com/vllm-project/vllm/blob/98dff2a81d747d1dba01a47f939f48c3526d4206/vllm/v1/worker/gpu/sample/bad_words.py
- https://github.com/vllm-project/vllm/pull/48824
- https://www.vulncheck.com/advisories/vllm-through-0.29.0-cross-request-logits-corruption-via-bad-words