GHSA-pc5q-2x89-w436LowCVSS 3.1
vLLM through 0.29.0 fails to properly validate bad_words token indices against the model's...
🔗 CVE IDs covered (1)
📋 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.
🔗 References (7)
- https://nvd.nist.gov/vuln/detail/CVE-2026-93989
- https://github.com/vllm-project/vllm/pull/48824
- 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://www.vulncheck.com/advisories/vllm-through-0.29.0-cross-request-logits-corruption-via-bad-words
- https://github.com/advisories/GHSA-pc5q-2x89-w436