vLLM 0.29.0 Bad_Words Index OOB CVE-2026-93989
CVE-2026-93989 Published on September 19, 2026

vLLM through 0.29.0 Cross-Request Logits Corruption via bad_words
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.

NVD

Vulnerability Analysis

CVE-2026-93989 can be exploited with network access, and requires small amount of user privileges. 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 no impact on confidentiality and integrity and availability.

Attack Vector:
NETWORK
Attack Complexity:
HIGH
Privileges Required:
LOW
User Interaction:
NONE

Weakness Type

What is an out-of-bounds array index Vulnerability?

The product uses untrusted input when calculating or using an array index, but the product does not validate or incorrectly validates the index to ensure the index references a valid position within the array.

CVE-2026-93989 has been classified to as an out-of-bounds array index vulnerability or weakness.


Products Associated with CVE-2026-93989

Want to know whenever a new CVE is published for Vllm? stack.watch will email you.

 

Affected Versions

vllm-project vllm: