CVE-2025-49840
Published: 15 July 2025
Summary
CVE-2025-49840 is a critical-severity Deserialization of Untrusted Data (CWE-502) vulnerability in Rvc-Boss Gpt-Sovits-Webui. Its CVSS base score is 9.8 (Critical).
Operationally, exploitation aligns with the MITRE ATT&CK technique Exploit Public-Facing Application (T1190); ranked in the top 42.9% of CVEs by exploit likelihood; it is not currently listed in the CISA KEV catalog; a public proof-of-concept is referenced.
This vulnerability is AI-related — categorised as APIs and Models; in the Other ATLAS/OWASP Terms risk domain.
The strongest mitigations our analysis identified are NIST 800-53 SI-10 (Information Input Validation) and SI-9 (Information Input Restrictions).
Threat & Defense at a Glance
Threat & Defense Details
Mitigating Controls (NIST 800-53 r5)AI
Validates the user-controlled gpt_path input before deserialization with torch.load, preventing processing of malicious serialized data.
Restricts GPT_dropdown inputs to whitelisted safe model paths, blocking arbitrary user-supplied paths leading to unsafe deserialization.
Implements memory protections like non-executable memory to mitigate remote code execution from deserialized malicious payloads.
MITRE ATT&CK Enterprise TechniquesAI
Why these techniques?
Unsafe deserialization via torch.load on user-controlled model path in webUI enables remote code execution, exploiting a public-facing web application.
NVD Description
GPT-SoVITS-WebUI is a voice conversion and text-to-speech webUI. In versions 20250228v3 and prior, there is an unsafe deserialization vulnerability in inference_webui.py. The GPT_dropdown variable takes user input and passes it to the change_gpt_weights function. In change_gpt_weights, the user input, here…
more
gpt_path is used to load a model with torch.load, leading to unsafe deserialization. At time of publication, no known patched versions are available.
Deeper analysisAI
CVE-2025-49840 is an unsafe deserialization vulnerability (CWE-502) in GPT-SoVITS-WebUI, a web interface for voice conversion and text-to-speech functionality within the RVC-Boss/GPT-SoVITS project. It affects versions 20250228v3 and prior, specifically in the inference_webui.py component. User input via the GPT_dropdown variable is passed to the change_gpt_weights function, which uses the controlled gpt_path parameter to load a model with torch.load, enabling arbitrary deserialization of untrusted data.
The vulnerability carries a CVSS v3.1 base score of 9.8 (AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:H/A:H), making it exploitable remotely by unauthenticated attackers with low attack complexity and no user interaction. Attackers can supply a malicious model path, leading to deserialization gadgets that compromise confidentiality, integrity, and availability, potentially resulting in remote code execution on the hosting server.
GitHub Security Lab advisories GHSL-2025-049 and GHSL-2025-053 detail the issue with code references in inference_webui.py (lines 310, 872, and 927), but at the time of publication on 2025-07-15, no known patched versions were available.
Details
- CWE(s)
Affected Products
AI Security AnalysisAI
- AI Category
- APIs and Models
- Risk Domain
- Other ATLAS/OWASP Terms
- OWASP Top 10 for LLMs 2025
- None mapped
- Classification Reason
- GPT-SoVITS-WebUI is an open-source web interface/platform for AI-based voice conversion and text-to-speech inference, utilizing PyTorch for model loading, which fits 'Other Platforms' as it is neither a framework, library, nor specific to NLP/CV/mobile/etc.