16 HIGH · THREAT ORANGE · 16 items · Generated in 732s
A security flaw in BerriAI litellm up to 1.82.2 allows remote manipulation, potentially leading to session expiration.
A weakness in BerriAI litellm up to 1.59.8 allows improper authentication, which can be exploited remotely.
A security vulnerability in BerriAI litellm up to 1.82.2 allows server-side request forgery, which can be exploited remotely.
A vulnerability in BerriAI litellm up to 1.82.2 allows remote execution of a manipulation that leads to missing authentication.
A vulnerability in BerriAI litellm up to 1.82.2 allows session expiration, exploitable remotely with publicly available exploits.
A security flaw in BerriAI litellm up to 1.82.5 allows remote manipulation of the prompt, leading to incorrect authorization.
LangChain framework has a vulnerability that allows disclosure of files outside the intended boundary when components receive path values from untrusted sources
Arbitrary code execution on the server via a malicious HuggingFace model.
vLLM's revision pinning controls do not consistently apply to all artifacts loaded for a model, allowing deployment of unpinned code.
Vulnerability in ASGI web servers and starlette's trust on those web servers enables authentication bypass of OpenAI API.
Integer truncation vulnerability in vLLM's GGUF dequantize kernels allows for information disclosure.
vulnerability to dependency confusion attack through a package on PyPI, allowing arbitrary code execution as root during Docker build and backdoor container ima
vLLM has a vulnerability that can lead to large memory usage, potentially causing denial of service.
vLLM has a vulnerability prior to version 0.23.1rc0 that can cause undefined behavior or CUDA errors, potentially crashing the inference worker.
A high-severity vulnerability was discovered in a large language model inference and serving engine, allowing an attacker to leak heap memory addresses through
Researchers found that language models can be confused by subtle changes in text format, leading to security vulnerabilities.