{"id":"CVE-2026-31219","aliases":["EUVD-2026-29503","GHSA-f5xg-pvfx-vwh9"],"status":"published","title":"The _load_model() function in the neural_magic_training.py script of the optimate project in...","description":"The _load_model() function in the neural_magic_training.py script of the optimate project in commit a6d302f912b481c94370811af6b11402f51d377f (2024-07-21) is vulnerable to insecure deserialization (CWE-502). When a user provides a single model file path (e.g., .pt or .pth) via the --model command-line argument, the function loads the file using torch.load() without enabling the weights_only=True security parameter. This allows the deserialization of arbitrary Python objects through the Pickle module. A remote attacker can exploit this by providing a maliciously crafted model file, leading to arbitrary code execution during deserialization on the victim's system.","severity":{"version":"3.1","score":8.8,"vector":"CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:U/C:H/I:H/A:H","level":"high","assessor":"adp","from":"cve"},"epss":{"day":"2026-11-11","score":0.00559,"percentile":0.47024,"model":"v2026.06.15"},"facts":{"in_kev":false,"known_exploited":false,"public_exploit":false,"fix_available":false},"cwes":["CWE-502"],"published":"2026-05-12T00:00:00Z","modified":"2026-10-28T02:31:02Z","references":[{"url":"https://github.com/nebuly-ai/optimate","tags":["package"]},{"url":"https://www.notion.so/CVE-2026-31219-35d1e139318881c8a629ec19671e74c3","tags":["web"]},{"url":"https://nvd.nist.gov/vuln/detail/CVE-2026-31219","tags":["advisory"]},{"url":"https://github.com/advisories/GHSA-f5xg-pvfx-vwh9"},{"url":"https://github.com/CVEProject/cvelistV5/tree/main/cves/2026/31xxx/CVE-2026-31219.json","tags":["advisory"]}],"sources":{"cve":[{"id":"CVE-2026-31219","description":"The _load_model() function in the neural_magic_training.py script of the optimate project in commit a6d302f912b481c94370811af6b11402f51d377f (2024-07-21) is vulnerable to insecure deserialization (CWE-502). When a user provides a single model file path (e.g., .pt or .pth) via the --model command-line argument, the function loads the file using torch.load() without enabling the weights_only=True security parameter. This allows the deserialization of arbitrary Python objects through the Pickle module. A remote attacker can exploit this by providing a maliciously crafted model file, leading to arbitrary code execution during deserialization on the victim's system.","status":"published","published":"2026-05-12T00:00:00Z","modified":"2026-05-15T14:01:46.092Z","cvss":[{"version":"3.1","score":8.8,"vector":"CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:U/C:H/I:H/A:H","level":"high","assessor":"adp"}],"cwes":["CWE-502"],"updated_at":"2026-11-02T14:19:50.92611Z"}],"euvd":[{"id":"EUVD-2026-29503","description":"The _load_model() function in the neural_magic_training.py script of the optimate project in commit a6d302f912b481c94370811af6b11402f51d377f (2024-07-21) is vulnerable to insecure deserialization (CWE-502). When a user provides a single model file path (e.g., .pt or .pth) via the --model command-line argument, the function loads the file using torch.load() without enabling the weights_only=True security parameter. This allows the deserialization of arbitrary Python objects through the Pickle module. A remote attacker can exploit this by providing a maliciously crafted model file, leading to arbitrary code execution during deserialization on the victim's system.","status":"published","published":"2026-05-12T00:00:00Z","modified":"2026-05-15T14:01:46Z","cvss":[{"version":"3.1","score":8.8,"vector":"CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:U/C:H/I:H/A:H","level":"high"}],"updated_at":"2026-11-02T19:04:28.872228Z"}],"ghsa":[{"id":"GHSA-f5xg-pvfx-vwh9","title":"The _load_model() function in the neural_magic_training.py script of the optimate project in...","description":"The _load_model() function in the neural_magic_training.py script of the optimate project in commit a6d302f912b481c94370811af6b11402f51d377f (2024-07-21) is vulnerable to insecure deserialization (CWE-502). When a user provides a single model file path (e.g., .pt or .pth) via the --model command-line argument, the function loads the file using torch.load() without enabling the weights_only=True security parameter. This allows the deserialization of arbitrary Python objects through the Pickle module. A remote attacker can exploit this by providing a maliciously crafted model file, leading to arbitrary code execution during deserialization on the victim's system.","status":"published","published":"2026-05-12T18:30:38Z","modified":"2026-05-15T15:31:36Z","cvss":[{"version":"3.1","score":8.8,"vector":"CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:U/C:H/I:H/A:H","level":"high"}],"cwes":["CWE-502"],"updated_at":"2026-11-02T15:08:10.185094Z"}],"nvd":[{"id":"CVE-2026-31219","description":"The _load_model() function in the neural_magic_training.py script of the optimate project in commit a6d302f912b481c94370811af6b11402f51d377f (2024-07-21) is vulnerable to insecure deserialization (CWE-502). When a user provides a single model file path (e.g., .pt or .pth) via the --model command-line argument, the function loads the file using torch.load() without enabling the weights_only=True security parameter. This allows the deserialization of arbitrary Python objects through the Pickle module. A remote attacker can exploit this by providing a maliciously crafted model file, leading to arbitrary code execution during deserialization on the victim's system.","status":"published","published":"2026-05-12T16:16:13.813Z","modified":"2026-06-17T10:33:27Z","cvss":[{"version":"3.1","score":8.8,"vector":"CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:U/C:H/I:H/A:H","level":"high","assessor":"secondary"}],"cwes":["CWE-502"],"updated_at":"2026-11-02T14:12:18.613259Z"}],"osv":[{"id":"CVE-2026-31219","description":"The _load_model() function in the neural_magic_training.py script of the optimate project in commit a6d302f912b481c94370811af6b11402f51d377f (2024-07-21) is vulnerable to insecure deserialization (CWE-502). When a user provides a single model file path (e.g., .pt or .pth) via the --model command-line argument, the function loads the file using torch.load() without enabling the weights_only=True security parameter. This allows the deserialization of arbitrary Python objects through the Pickle module. A remote attacker can exploit this by providing a maliciously crafted model file, leading to arbitrary code execution during deserialization on the victim's system.","status":"published","published":"2026-05-12T00:00:00Z","modified":"2026-10-28T02:31:02.790070944Z","cvss":[{"version":"3.1","score":8.8,"vector":"CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:U/C:H/I:H/A:H","level":"high"}],"updated_at":"2026-11-02T15:45:00.861953Z"}]}}