CVE-2026-5585: Tencent AI-Infra-Guard information disclosure in Task Detail Endpoint

CVE-2026-5585 · Severity: medium · CVSS 5.3 · Published 2026-04-05

Tencent AI-Infra-Guard is a tool used for managing AI infrastructure and tasks. A security flaw in the task management component allows unauthorized individuals to view sensitive information, specifically plaintext API tokens for AI models like GPT-4. This could lead to the theft of expensive API credits, unauthorized use of AI services, and potential financial loss for the organization.

Technologies: Tencent Ai-Infra-Guard. Vendors: Tencent.

Quick answers

Is CVE-2026-5585 exploited in the wild?
No. Junglewise Threat Intelligence has no report of CVE-2026-5585 being exploited in the wild.
How severe is CVE-2026-5585?
Medium severity, with a CVSS score of 5.3, published 5 April 2026.
Which products and versions are affected?
Tencent Ai-Infra-Guard 4.0.
Is there a fix for CVE-2026-5585?
No fixed version is recorded yet. Check the vendor's advisory in the references below. If you run it, Junglewise drafts a mitigation that fits your setup and applies it once you approve, with senior security engineers behind it.
How is CVE-2026-5585 attacked?
It is attacked over the network, in a software library.
Am I affected by CVE-2026-5585?
Check whether you run Tencent Ai-Infra-Guard at an affected version. Junglewise does that check for you around the clock: connect your code, cloud and domains, and it matches new vulnerabilities like this one against what you actually run and tells you when one reaches you.

Technical details

A vulnerability exists in the Task Detail API endpoint (/api/v1/app/tasks/{sessionId}) of Tencent AI-Infra-Guard 4.0 due to missing data masking and improper authentication. The root cause is located in common/websocket/task_manager.go, where the GetTaskDetail function returns the session.Params field—containing raw API tokens—without invoking masking functions. Additionally, the setupIdentityMiddleware in common/websocket/server.go fails to perform cryptographic verification, blindly trusting HTTP headers. An unauthenticated remote attacker can exploit this by guessing or obtaining a session ID to retrieve another user's plaintext AI model credentials. As of the advisory date, no patch has been released by the vendor.

Affected products

  • Tencent AI-Infra-Guard 4.0

Timeline

  • 2026-03-20: disclosed: Public disclosure via GitHub Gist PoC
  • 2026-04-05: advisory: Vulnerability published by VulDB/NVD

References

This page as Markdown. Free to reuse under CC BY 4.0 with a link back.