---
id: "CVE-2024-49375"
page: "https://infrastack.ai/live/CVE-2024-49375"
kind: "vulnerability"
aliases: ["CVE-2024-49375", "GHSA-cpv4-ggrr-7j9v"]
title: "Rasa Allows Remote Code Execution via Remote Model Loading"
severity: "critical"
cvss: 9.0
epss: 0.00937
exploited: false
status: "open"
category: "package"
products: [{"name": "rasa-pro", "fixed": ["3.10.12"], "affected": ">= 3.10.0, < 3.10.12", "ecosystem": "pypi"}, {"name": "rasa-pro", "fixed": ["3.9.16"], "affected": ">= 3.9.0, < 3.9.16", "ecosystem": "pypi"}, {"name": "rasa-pro", "fixed": ["3.8.18"], "affected": "< 3.8.18", "ecosystem": "pypi"}, {"name": "rasa", "fixed": ["3.6.21"], "affected": "< 3.6.21", "ecosystem": "pypi"}]
ecosystems: ["pypi"]
tags: ["cwe-502", "cwe-94", "ml-chatbots", "package", "python", "rasa"]
not_affected_if: ["api-disabled", "auth-configured", "default-config"]
published_at: "2025-01-14T16:00:21+00:00"
first_seen_at: "2026-11-16T16:18:39+00:00"
updated_at: "2026-11-16T16:19:19+00:00"
sources: [{"source": "ghsa", "url": "https://github.com/advisories/GHSA-cpv4-ggrr-7j9v", "first_seen_at": "2026-11-16T16:18:39+00:00"}, {"source": "nvd", "url": "https://nvd.nist.gov/vuln/detail/CVE-2024-49375", "first_seen_at": "2026-11-16T16:38:06+00:00"}]
exploit_sources: [{"type": "epss", "score": 0.00937, "url": "https://api.first.org/data/v1/epss?cve=CVE-2024-49375"}]
---

# Rasa Allows Remote Code Execution via Remote Model Loading

Link for people: https://infrastack.ai/live/CVE-2024-49375 (this .md is for agents)

Rasa Pro and Rasa Open Source let a model be loaded remotely through the HTTP API, and a maliciously crafted model can execute code on the instance. An attacker who can load a model gets remote code execution, and with the API open and no authentication this needs no login.

_Summary, check and fix written by a model from the sources; labelled as such._

## Affected

- pypi:rasa-pro affected >= 3.10.0, < 3.10.12 fixed 3.10.12
- pypi:rasa-pro affected >= 3.9.0, < 3.9.16 fixed 3.9.16
- pypi:rasa-pro affected < 3.8.18 fixed 3.8.18
- pypi:rasa affected < 3.6.21 fixed 3.6.21

## How to check

- Check the installed version with `pip show rasa-pro` or `pip show rasa`.
- Check whether the HTTP API is enabled, for example by looking for `--enable-api` in startup commands or service files.
- Check whether authentication is configured for the Rasa HTTP API.

## Fix

- Rasa Pro: upgrade to `3.8.18`, `3.9.16` or `3.10.12` depending on the branch in use.
- Rasa Open Source: upgrade to `3.6.21`.
- After upgrading, retrain the model with the fixed version.

## Timeline

- 2026-11-16T16:18:39+00:00 ghsa https://github.com/advisories/GHSA-cpv4-ggrr-7j9v
- 2026-11-16T16:38:06+00:00 nvd https://nvd.nist.gov/vuln/detail/CVE-2024-49375

## Documents

- [advisory] Maintainer's advisory: api.github.com https://api.github.com/repos/RasaHQ/security-advisories/security-advisories/GHSA-cpv4-ggrr-7j9v
- [advisory] Vendor advisory: github.com https://github.com/RasaHQ/rasa-pro-security-advisories/security/advisories/GHSA-cpv4-ggrr-7j9v
- [fix] Patch: github.com https://github.com/RasaHQ/rasa/commit/2bb1d779d4f4acaf70b6dfa35dd1899dccbb1ae6

## From the sources

## Vulnerability
A vulnerability has been identified in Rasa Pro and Rasa Open Source that enables an attacker who has the ability to load a maliciously crafted model remotely into a Rasa instance to achieve Remote Code Execution.

The prerequisites for this are:
- The HTTP API must be enabled on the Rasa instance eg with `--enable-api`. This is not the default configuration.
- For unauthenticated RCE to be exploitable, the user must not have configured any authentication or other security controls recommended in our documentation.
- For authenticated RCE, the attacker must posses a valid authentication token or JWT to interact with the Rasa API.

## Fix
We encourage you to upgrade to a version of Rasa that includes a fix. These are:

- Rasa Pro 3.8.18, 3.9.16, 3.10.12
- Rasa Open Source 3.6.21

Once you have upgraded your Rasa Pro or Open Source installation, you will need to retrain your model using the fixed version of Rasa Pro or Open Source. If you have a custom component that inherits from one of the components listed below and modified the persist or load method, make sure to update your code. Please contact us in case you encounter any problems.

Affected components:

- `CountVectorFeaturizer`
- `LexicalSyntacticFeaturizer`
- `LogisticRegressionClassifier`
- `SklearnIntentClassifier`
- `DIETClassifier`
- `CRFEntityExtractor`
- `TrackerFeaturizer`
- `TEDPolicy`
- `UnexpectedIntentTEDPolicy`

If you are unable to upgrade immediately, please follow our mitigation advice below.

## Mitigation Advice

- The unauthenticated RCE can be mitigated by ensuring that you enable authentication for the Rasa HTTP API if this is enabled. This means that in addition to passing `--enable-api` to Rasa, you should also pass an authentication configuration as described in our documentation [here](https://rasa.com/docs/rasa-pro/production/rest-api/#enabling-the-rest-api). 

- The authenticated RCE can be mitigated by:
    - Ensuring that you only load models from trusted sources such as your own CI pipelines. You can check file hashes for model archives to confirm that a model has not been tampered with between training and use.
    - Confirming that you have sufficient access controls and utilize the principle of least privilege to control who in your organization has the ability to interact with the Rasa API even with authentication enabled.

## Future Releases
As an additional security step, a future release of Rasa Pro will remove the ability to enable the API without any authentication method enabled.

## Credit
Rasa would like to thank Julian Scheid from Deutsche Telekom Security GmbH for responsible disclosure of this vulnerability.

---

_Free to read, with fair use of about 60 pages a minute. For higher limits or API access: sales@infrastack.ai_
