CVE-2026-63766

GPT-SoVITS 20250606v2pro OS Command Injection via webui.py

Published Jul 20, 2026 CVSS 4.0 · 9.3 Critical EPSS 2.1% KEV · Not listed
Description
GPT-SoVITS through 20250606v2pro contains an OS command injection vulnerability in webui.py where ASR, slice, denoise, and uvr5 functions interpolate unsanitized Gradio textbox values directly into shell commands executed with shell=True. Attackers can inject shell metacharacters through path parameters to execute arbitrary OS commands as the server process user without authentication.

Analysis

AI-generated

Technical impact

GPT-SoVITS version 20250606v2pro includes an OS command injection flaw located in the webui.py module; functions such as ASR, slice, denoise, and uvr5 interpolate raw Gradio textbox input directly into shell commands executed with shell=True. The flaw is classified as CWE-78 and allows an attacker to inject shell metacharacters through path parameters, causing execution as the server process user and enabling remote code execution without authentication.

Affected Systems

The vulnerable implementation is present in the RVC-Boss GPT-SoVITS release 20250606v2pro; any deployment running this exact version is affected. The likely earlier releases might also contain this flaw if they include the same webui.py code path, but only 20250606v2pro is confirmed to be vulnerable.

Risk and Exploitability

The CVSS score of 9.3 reflects critical severity, while the EPSS score of 2% indicates a low but non‑zero probability of exploitation. The vulnerability is not yet listed in the CISA KEV catalog, suggesting no known large‑scale attacks, but the lack of authentication and the ability to trigger the flaw remotely make it highly exploitable. Immediate remediation is warranted due to the potential for complete system compromise.

Remediation

No solution or workaround provided in the CVE record.

OpenCVE Recommended Actions

  • Upgrade to a patched release of GPT-SoVITS that sanitizes all Gradio textbox input before embedding it in shell commands.
  • If an immediate upgrade cannot be performed, implement a server‑side filter that escapes or removes shell metacharacters from all user‑supplied path parameters before they reach webui.py.
  • Run the GPT-SoVITS application under the least privileged system account and, if possible, isolate its subprocesses in a sandboxed container to limit damage from any successful injection.

Generated by OpenCVE AI on July 30, 2026 at 18:37 UTC.

Tracking

Sign in to view the affected projects.

Advisories

No advisories available
History
Type Values Removed Values Added
First Time appeared Rvc-boss
Rvc-boss gpt-sovits-webui
CPEs cpe:2.3:a:rvc-boss:gpt-sovits-webui:*:*:*:*:*:*:*:*
Vendors & Products Rvc-boss
Rvc-boss gpt-sovits-webui
Type Values Removed Values Added
Metrics ssvc

{'options': {'Automatable': 'yes', 'Exploitation': 'poc', 'Technical Impact': 'total'}, 'version': '2.0.3'}

Type Values Removed Values Added
Description GPT-SoVITS through 20250606v2pro contains an OS command injection vulnerability in webui.py where ASR, slice, denoise, and uvr5 functions interpolate unsanitized Gradio textbox values directly into shell commands executed with shell=True. Attackers can inject shell metacharacters through path parameters to execute arbitrary OS commands as the server process user without authentication.
Title GPT-SoVITS 20250606v2pro OS Command Injection via webui.py
Weaknesses CWE-78
References
Metrics cvssV3_1

{'score': 9.8, 'vector': 'CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:H/A:H'}

cvssV4_0

{'score': 9.3, 'vector': 'CVSS:4.0/AV:N/AC:L/AT:N/PR:N/UI:N/VC:H/VI:H/VA:H/SC:N/SI:N/SA:N'}

Weaknesses
  • CWE-78

    Improper Neutralization of Special Elements used in an OS Command ('OS Command Injection')

Sources

Status
PUBLISHED
Assigner
VulnCheck
Published
Updated
Jul 28, 2026, 1:05 AM
Reserved
Jul 18, 2026, 12:34 PM
Status
Deferred
Published
Jul 20, 2026, 8:16 PM
Modified
Oct 24, 2026, 8:10 PM