Executive Summary

CVE-2026-5585 is a vulnerability in Tencent AI-Infra-Guard 4.0 that causes sensitive information disclosure through the Task Detail API endpoint. Specifically, when users submit tasks with sensitive API tokens, these tokens are stored in plaintext and shared by default. The API endpoint that retrieves task details does not mask these tokens, exposing them in the response.

Additionally, the authentication mechanism is critically flawed, relying only on an arbitrary username HTTP header without proper verification. This allows any unauthenticated attacker to impersonate users and access sensitive API tokens belonging to others.

Useful 0 Not Useful 0

Impact Analysis

This vulnerability allows unauthenticated attackers to obtain high-value API credentials for AI models belonging to other users or the platform. Such exposure can lead to severe financial loss due to unauthorized usage of API quotas and potential cross-platform compromise if tokens are reused elsewhere.

• Unauthorized access to sensitive API tokens.
• Potential financial loss from token theft and misuse.
• Risk of broader compromise if tokens are reused across platforms.

Useful 0 Not Useful 0

Detection Guidance

This vulnerability can be detected by attempting to access the Task Detail API endpoint to see if sensitive API tokens are exposed without authentication.

A practical detection method involves using curl commands to simulate both a victim user submitting a task with a sensitive API token and an unauthenticated attacker retrieving the task details.

• Submit a task with a sensitive API token using a POST request: curl -X POST http://<target>:8088/api/v1/app/taskapi/tasks -d '{"params": {"api_token": "sensitive_token_here"}}' -H 'Content-Type: application/json'
• Extract the session_id from the response.
• Retrieve the task details without authentication using a GET request: curl http://<target>:8088/api/v1/app/tasks/{session_id}

If the response includes the plaintext API token, the vulnerability is present.

Useful 0 Not Useful 0

Mitigation Strategies

Currently, no patches are available for this vulnerability.

Immediate mitigation steps include restricting access to the AI-Infra-Guard service, especially the Task Detail API endpoint, to trusted and authenticated users only.

Implement network-level controls such as firewall rules or VPN access to limit exposure.

Avoid sharing sensitive API tokens via the Task Detail endpoint until a fix is released.

Monitor logs for unauthorized access attempts to the Task Detail API.

Consider rotating any exposed API tokens immediately to prevent misuse.

Useful 0 Not Useful 0

Compliance Impact

This vulnerability results in the exposure of sensitive API tokens to unauthenticated attackers due to lack of token masking and weak authentication controls.

Such unauthorized disclosure of sensitive information can lead to violations of data protection regulations like GDPR and HIPAA, which require strict controls to protect sensitive data and prevent unauthorized access.

Useful 0 Not Useful 0