Is CVE-2026-123019 exploited in the wild?

No. Junglewise Threat Intelligence has no report of CVE-2026-123019 being exploited in the wild.

How severe is CVE-2026-123019?

High severity, with a CVSS score of 7.3, published 5 November 2026.

Which products and versions are affected?

Eosphoros-Ai DB-GPT up to 0.8.2.

How is CVE-2026-123019 attacked?

It is attacked over the network.

Am I affected by CVE-2026-123019?

Check whether you run Eosphoros-Ai DB-GPT 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.

CVE-2026-123019: eosphoros-ai DB-GPT remote code execution in HuggingFace adapter

CVE-2026-123019 · Severity: high · CVSS 7.3 · Published 2026-11-05

DB-GPT is an AI data assistant that integrates machine learning models. Versions up to 0.8.2 allow unauthenticated attackers to execute arbitrary code on the server by registering a malicious HuggingFace model, because the model loading endpoint lacks authentication and automatically trusts remote model code by default. This grants attackers complete control over the server process, allowing them to steal credentials, modify data, or disrupt service.

Technologies: Eosphoros-Ai DB-GPT. Vendors: Eosphoros-Ai.

Quick answers

Is CVE-2026-123019 exploited in the wild?: No. Junglewise Threat Intelligence has no report of CVE-2026-123019 being exploited in the wild.
How severe is CVE-2026-123019?: High severity, with a CVSS score of 7.3, published 5 November 2026.
Which products and versions are affected?: Eosphoros-Ai DB-GPT up to 0.8.2.
Is there a fix for CVE-2026-123019?: No fixed version is recorded yet. Restrict network access to the model API endpoint (default port 5670) to trusted clients only, or disable remote model loading via the HuggingFace adapter until an upgrade is available. Configure an API authentication requirement before deploying to network-reachable infrastructure." 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-2026-123019 attacked?: It is attacked over the network.
Am I affected by CVE-2026-123019?: Check whether you run Eosphoros-Ai DB-GPT 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.

How to fix

Restrict network access to the model API endpoint (default port 5670) to trusted clients only, or disable remote model loading via the HuggingFace adapter until an upgrade is available. Configure an API authentication requirement before deploying to network-reachable infrastructure."

Technical details

The vulnerability combines CWE-829 (inclusion of untrusted functionality) and CWE-306 (missing authentication): the HuggingFace adapter defaults trust_remote_code=True, causing transformers.from_pretrained() to execute arbitrary Python code from model repositories without sandbox restrictions; simultaneously, the model registration endpoint POST /api/v2/serve/model/models requires no authentication. An unauthenticated attacker can register a malicious model and trigger code execution in the server process with full privileges. A public proof-of-concept has been published.

Affected products

eosphoros-ai DB-GPT up to 0.8.2

Timeline

2026-10-14: disclosed: Reported via GitHub issue #3290
2026-11-05: advisory

References

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