The vulnerability in Red Hat OpenShift AI (RHOAI) MaaS Gateway arises from an insecure default configuration of allowedRoutes.namespaces.from, which permits any user in an authorized namespace to hijack shared model-serving traffic. This includes interception and alteration of sensitive information such as access keys, input prompts, and outputs. The issue is classified as improper access control (CWE-284). While restricting Gateway access to certain namespaces can mitigate risk, it does not prevent hijacking among authorized namespaces. Fully locking down access is a trade-off against the component's self-service design. The vulnerability has a CVSS v3.1 score of 8.8 (high severity) with network attack vector, low complexity, low privileges required, and no user interaction needed. No vendor-provided patch or official fix is currently available according to the advisory.

An attacker with low privileges in any authorized namespace can intercept, read, log, and modify all model-serving traffic on the MaaS Gateway. This leads to significant confidentiality breaches (exposure of access keys and sensitive data), integrity violations (data tampering), and availability impact. The flaw compromises the security boundaries between namespaces, allowing unauthorized data access and manipulation within the shared model-serving environment.

Currently, no official patch or fix is available. Mitigation involves restricting Gateway access to a limited set of namespaces; however, this does not prevent traffic hijacking among authorized namespaces. Fully locking down access to the MaaS Gateway can prevent exploitation but disables the self-service functionality of the component. Users should monitor the Red Hat advisory for updates on fixes and consider access restrictions as an interim measure.