Description

HCL AION is affected by a missing encryption vulnerability in a second WebSocket connection defined in the GenAI web interface. The connection used the unencrypted WebSocket protocol (ws://test) instead of an encrypted (wss://) channel. This may allow data exchanged over the connection to be intercepted or read by an attacker on the network, potentially resulting in unintended behavior or security impact under certain conditions.

Executive Summary

Technical Details

Mitigations

References

Additional Metadata

{
  "score": "3.7",
  "pubDate": "2026-11-06T14:17:13.407Z",
  "pubdate": "2026-11-06T14:17:13.407Z",
  "executiveSummary": "HCL AION contains a critical security deficiency involving the implementation of a secondary WebSocket connection within the GenAI web interface. The application utilizes an insecure, unencrypted protocol (ws://) for data transmission rather than the industry-standard encrypted (wss://) protocol. This missing encryption constitutes a CWE-311 (Failure to Encrypt Sensitive Data) vulnerability, exposing transit data to unauthorized interception.\nThe vulnerability resides within the network communication layer of the GenAI interface, potentially allowing local or remote network attackers to perform Man-in-the-Middle (MitM) attacks. By eavesdropping on the unencrypted channel, an attacker can capture sensitive information exchanged between the client and the server, leading to potential data exposure, session manipulation, or unauthorized access to sensitive GenAI operations. The risk is significant for deployments operating within environments where network traffic is not otherwise secured via underlying infrastructure controls. No complex authentication or elevated privileges are required to intercept this traffic, provided the attacker maintains a position on the network path between the client and the AION instance.",
  "technicalDetails": "The vulnerability stems from an insecure implementation of the WebSocket handshake and transport protocol within the HCL AION GenAI web interface. Specifically, a secondary WebSocket connection is hardcoded or configured to initiate via the 'ws://' URI scheme, which transmits data in plaintext without applying Transport Layer Security (TLS). This is a violation of secure communication best practices for web applications, as it bypasses the confidentiality guarantees provided by the WSS (WebSocket Secure) protocol.\nThe attack flow begins when an attacker positioned on the same network segment as the victim—such as an open Wi-Fi network, an compromised gateway, or an internal enterprise subnet—monitors traffic directed at the HCL AION instance. Because the handshake and subsequent data frames are transmitted in the clear, the attacker can leverage standard packet capture tools or proxying software to reconstruct the WebSocket stream. The absence of encryption means the protocol overhead, including headers and the full payload of the messages, is visible to the interceptor.\nThe technical impact is multifaceted. Firstly, it allows for 'eavesdropping' where sensitive queries, model outputs, or internal system state information passed through the WebSocket can be parsed by the attacker. Secondly, because WebSockets are stateful and often used for bidirectional communication in AI interfaces, a MitM attacker can potentially perform 'injection' or 'tampering' if the interface does not enforce strict message integrity or authentication tokens within the encrypted layer. Even in scenarios where the primary interface uses HTTPS, the secondary unencrypted connection creates a 'split-trust' scenario where auxiliary functions remain vulnerable.\nExploitation requires no special authentication to the application itself; it relies solely on the ability of the attacker to see the traffic. By intercepting the initial HTTP upgrade request, the attacker can verify that the connection is destined for a non-TLS WebSocket endpoint. Once the connection is established, any sensitive tokens, user data, or application-specific commands transmitted over this secondary channel are exposed in plaintext. The post-exploitation impact includes the potential for data exfiltration, the compromising of user session integrity, and the leakage of proprietary logic or data processed by the GenAI engine. The vulnerability persists as long as the application code requests an insecure 'ws://' connection rather than enforcing 'wss://' at the application layer."
}