HighVulnerability
CVE-2026-134172: audio-separator insecure deserialization of malicious Demucs model files
- Source
- NVD(opens in a new tab)
- Identifier
- CVE-2026-134172
- Published
- Record updated
- Known exploitation
- Not listed in the CISA Known Exploited Vulnerabilities catalog at the last check.
Summary
audio-separator versions before 0.47.2 contain an insecure deserialization flaw. Malicious Demucs .th model files loaded via torch.load with weights_only=False can execute code when a victim runs Demucs separation. Local users able to write to the shared /tmp/audio-separator-models/ default directory, or attackers controlling unverified model download sources, can exploit it.
Mitigation
Fixed in 0.47.2 (audio-separator versions before 0.47.2 are affected).
Related items
- MediumCVE-2026-133836: hankcs HanLP deserialization via model_file in RNNLanguageModelSimilar attack · NVD/CVE Database
- MediumCVE-2026-128136: NLTK pickle loading lets crafted models read or truncate local filesSimilar attack · NVD/CVE Database
- HighCVE-2026-76818: Axon rtkx bootstrap runs unverified downloaded binarySimilar attack · NVD/CVE Database
- MediumMicrosoft’s Agent Framework PyPI package compromised to steal credentialsSimilar attack · BleepingComputer
- InfoOX Security Launches OX Radar and Response Center to Show Companies Which Cyber Threats Actually Affect ThemSimilar attack · CSO Online