CVE-2025-32377: 
Python vulnerability analysis and mitigation

Rasa Pro, a framework for building scalable, dynamic conversational AI assistants that integrate large language models (LLMs), was found to contain a vulnerability where voice connectors failed to properly implement authentication even when a token was configured in the credentials.yml file. The vulnerability was identified and tracked as CVE-2025-32377, affecting versions up to 3.12.5, 3.11.6, 3.10.18, and 3.9.17. The issue was discovered and disclosed on April 17, 2025 (GitHub Advisory).

The vulnerability specifically impacts the audiocodes_stream, genesys, and jambonz connectors in Rasa Pro. The issue stems from improper implementation of authentication mechanisms in voice connectors, even when authentication tokens are explicitly configured in the credentials.yml file. The vulnerability has been assigned a CVSS v3.1 base score of 6.5 (Medium) with a vector string of CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:L/A:L, indicating network accessibility with low attack complexity and no required privileges or user interaction (GitHub Advisory).

The vulnerability allows attackers to submit voice data to the Rasa Pro assistant from unauthenticated sources, potentially compromising the integrity and availability of the system. The impact is particularly significant for production environments using the affected voice connectors (GitHub Advisory).

The vulnerability has been patched in Rasa Pro versions 3.9.20, 3.10.19, 3.11.7, and 3.12.6. Users are strongly recommended to upgrade to these fixed versions. For connectors where authentication is not supported (such as Twilio), users should implement additional compensating controls (GitHub Advisory).