Description

A flaw was found in the TrustyAI Service (TAS) deployment. This vulnerability allows any pod on the cluster network to bypass authentication and directly access the TAS backend API. An attacker can exploit this to read, tamper with, or delete monitoring data and configurations, and inject arbitrary data into the service, potentially disrupting tenant operations.

Executive Summary

Technical Details

Mitigations

References

Additional Metadata

{
  "score": "8.0",
  "pubDate": "2026-08-10T21:17:19.743Z",
  "pubdate": "2026-08-10T21:17:19.743Z",
  "executiveSummary": "A critical authentication bypass vulnerability has been identified within the TrustyAI Service (TAS) deployment architecture. The security flaw permits arbitrary pods residing on the cluster network to circumvent authentication controls entirely and establish direct communication with the TAS backend API. The primary vulnerability type is an authentication bypass resulting from improper network access controls or API security enforcement. The potential impact of this security deficiency is severe, enabling unauthorized threat actors or compromised cluster workloads to read, tamper with, or delete critical monitoring data and configuration states. Furthermore, attackers can exploit this exposure to inject arbitrary data directly into the service, thereby disrupting tenant operations and compromising the integrity of AI/ML monitoring pipelines. The affected product is the TrustyAI Service (TAS) deployment. Risk implications include complete loss of data confidentiality, integrity, and availability for the affected monitoring service. Attacker capabilities encompass full interaction with the backend API without possessing valid credentials, provided the actor has initial code execution within any pod on the cluster network. Exploitation requirements involve network reachability to the TAS backend API within the cluster topology and the absence of robust network policies or service mesh authentication boundaries.",
  "technicalDetails": "The root cause of the vulnerability lies in the inadequate enforcement of authentication and authorization mechanisms at the TrustyAI Service (TAS) backend API boundary. Specifically, the service deployment fails to validate incoming HTTP requests or enforce mutual TLS (mTLS) and token-based authentication for connections originating from within the internal cluster network. Consequently, the vulnerable component, the TAS backend API, trusts requests arriving from any pod on the cluster network implicitly.\nThe network exposure of the TAS backend API is overly broad, allowing internal cluster traffic to reach administrative or operational endpoints without traversing a secure API gateway or sidecar proxy configured with strict access controls. Because authentication requirements are absent or improperly configured for internal cluster communication routes, no privilege requirements or valid credentials are needed to interact with the backend service once network proximity to the TAS pod or service endpoint is achieved.\nThe step-by-step attack flow proceeds as follows: First, an attacker achieves initial execution within any arbitrary pod on the same Kubernetes or container orchestration cluster, either through a separate application vulnerability, a compromised dependency, or insider access. Second, the attacker leverages internal cluster DNS or service discovery mechanisms to resolve the TrustyAI Service (TAS) backend API endpoint. Third, the attacker issues direct HTTP or HTTPS requests to the TAS backend API, bypassing any perimeter security controls that might protect external routes. Fourth, because the TAS backend API lacks internal authentication enforcement, the requests are processed as legitimate. Finally, the attacker executes unauthorized payload behaviors, such as reading sensitive monitoring data, deleting or tampering with configurations, and injecting malicious or arbitrary data into the service to disrupt tenant operations."
}