Логотип exploitDog
Консоль
Логотип exploitDog

exploitDog

redhat логотип

CVE-2026-45833

Опубликовано: 12 июн. 2026
Источник: redhat
CVSS3: 7.5
EPSS Низкий

Описание

A code injection vulnerability in version 0.4.17 or later of the ChromaDB Python project allows an authenticated attacker to run arbitrary code on the server by sending a malicious model repository and trust_remote_code set to true in the /api/v2/tenants/default_tenant/databases/default_database/collections/{collection_id} if they have the UPDATE_COLLECTION permission.

A flaw was found in the ChromaDB Python project. An authenticated attacker with UPDATE_COLLECTION permission could exploit a code injection vulnerability. By sending a malicious model repository to a specific API endpoint with trust_remote_code enabled, the attacker can execute arbitrary code on the server. This could lead to a complete compromise of the affected system.

Отчет

This is a post-authentication ChromaDB FastAPI flaw: exploitation requires a valid user with UPDATE_COLLECTION and a crafted request that sets trust_remote_code: true on a malicious HuggingFace model. It is not unauthenticated remote code execution. RH AI products ship a vulnerable chromadb version but do not expose the Chroma FastAPI API as the default product surface. RHOAI AutoRAG uses Llama Stack with Milvus/pgvector; RHEL AI bootc images include chromadb as a library dependency, not a network-facing Chroma server. Without a customer or misconfiguration that runs and exposes the Chroma Python server to untrusted users, the upstream attack path is not reachable. Upstream CVSS 9.4 Critical assumes network access to Chroma’s API with exploitable permissions. That does not match default RHOAI/RHEL AI architecture, so Important is the appropriate RH impact.

Меры по смягчению последствий

To mitigate this issue, ensure that the trust_remote_code setting in ChromaDB is disabled. This setting prevents the execution of arbitrary code from remote model repositories. Consult ChromaDB documentation for specific configuration instructions to disable trust_remote_code. If the ChromaDB service is restarted or reloaded, verify that the setting remains disabled.

Затронутые пакеты

ПлатформаПакетСостояниеРекомендацияРелиз
Red Hat Enterprise Linux AI (RHEL AI) 3rhelai3/bootc-cuda-rhel9Will not fix
Red Hat Enterprise Linux AI (RHEL AI) 3rhelai3/bootc-gaudi-rhel9Will not fix
Red Hat Enterprise Linux AI (RHEL AI) 3rhelai3/bootc-rocm-rhel9Will not fix
Red Hat Enterprise Linux AI (RHEL AI) 3rhelai3/disk-image-cuda-rhel9Will not fix
Red Hat OpenShift AI (RHOAI)rhoai/odh-autorag-rhel9Not affected

Показывать по

Дополнительная информация

Статус:

Important
Дефект:
CWE-94
https://bugzilla.redhat.com/show_bug.cgi?id=2488430chromadb: ChromaDB: Arbitrary Code Execution via Code Injection

EPSS

Процентиль: 27%
0.00342
Низкий

7.5 High

CVSS3

Связанные уязвимости

CVSS3: 8.8
nvd
около 2 месяцев назад

A code injection vulnerability in version 0.4.17 or later of the ChromaDB Python project allows an authenticated attacker to run arbitrary code on the server by sending a malicious model repository and trust_remote_code set to true in the /api/v2/tenants/default_tenant/databases/default_database/collections/{collection_id} if they have the UPDATE_COLLECTION permission.

CVSS3: 8.8
github
около 2 месяцев назад

A code injection vulnerability in version 0.4.17 or later of the ChromaDB Python project allows an authenticated attacker to run arbitrary code on the server by sending a malicious model repository and trust_remote_code set to true in the /api/v2/tenants/default_tenant/databases/default_database/collections/{collection_id} if they have the UPDATE_COLLECTION permission.

EPSS

Процентиль: 27%
0.00342
Низкий

7.5 High

CVSS3