{"id":"CVE-2026-47117","aliases":["EUVD-2026-33942","GHSA-m3v4-v5gx-7wf5","PYSEC-2026-2852"],"status":"published","title":"OpenMed \u003c 1.5.2 Remote Code Execution via PII Model Loading","description":"OpenMed before 1.5.2 contains a remote code execution vulnerability in the PII privacy-filter model loading path. The privacy-filter dispatcher used broad substring matching on the user-supplied model_name parameter, allowing a value such as attacker/foo-privacy-filter-bar to route through a path that loads Hugging Face models with trust_remote_code=True. An unauthenticated attacker can supply a malicious model repository containing custom Transformers code via auto_map in config.json or tokenizer_config.json, which is imported and executed with the privileges of the OpenMed service process.","severity":{"version":"4.0","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","level":"critical","assessor":"cna","from":"cve"},"epss":{"day":"2026-11-10","score":0.00915,"percentile":0.60442,"model":"v2026.06.15"},"facts":{"in_kev":false,"known_exploited":false,"public_exploit":true,"fix_available":true},"cwes":["CWE-94"],"published":"2026-06-02T14:15:17Z","modified":"2026-11-09T02:48:04Z","exploits":[{"source":"pocgithub","id":"SaiTeja-Erukude/CVE-2026-47117-openmed-rce","title":"SaiTeja-Erukude/CVE-2026-47117-openmed-rce","date":"2026-08-15","url":"https://github.com/SaiTeja-Erukude/CVE-2026-47117-openmed-rce"}],"fixes":[{"ecosystem":"PyPI","package":"openmed","fixed_in":["1.5.2"],"sources":["ghsa","osv"]}],"affected":[{"source":"ghsa","ecosystem":"PyPI","package":"openmed","ranges":[{"type":"ECOSYSTEM","events":[{"introduced":"0"},{"fixed":"1.5.2"}]}]},{"source":"osv","ecosystem":"PyPI","package":"openmed","ranges":[{"type":"ECOSYSTEM","events":[{"introduced":"0"},{"fixed":"1.5.2"}]}],"versions":["0.1","0.1.1","0.1.10","0.1.10rc1","0.1.4","0.1.5","0.1.7","0.1.8","0.1.8rc0","0.1.8rc2","0.1.9","0.1.9rc1","0.2.0","0.2.0rc1","0.2.0rc2","0.2.1","0.2.2","0.3.0","0.4.0","0.5.0","0.5.1","0.5.5","0.5.6","0.5.7","0.5.8","0.6.0","0.6.1","0.6.2","0.6.3","0.6.4","1.0.0","1.1.0","1.2.0","1.3.0","1.4.0","1.5.0","1.5.1"]},{"source":"osv","ecosystem":"PyPI","package":"openmed","ranges":[{"type":"ECOSYSTEM","events":[{"introduced":"0"},{"fixed":"1.5.2"}]}],"versions":["0.1","0.1.1","0.1.10","0.1.10rc1","0.1.4","0.1.5","0.1.7","0.1.8","0.1.8rc0","0.1.8rc2","0.1.9","0.1.9rc1","0.2.0","0.2.0rc1","0.2.0rc2","0.2.1","0.2.2","0.3.0","0.4.0","0.5.0","0.5.1","0.5.5","0.5.6","0.5.7","0.5.8","0.6.0","0.6.1","0.6.2","0.6.3","0.6.4","1.0.0","1.1.0","1.2.0","1.3.0","1.4.0","1.5.0","1.5.1"]}],"products":[{"vendor":"maziyarpanahi","product":"openmed"}],"references":[{"url":"https://github.com/SaiTeja-Erukude/CVE-2026-47117-openmed-rce","tags":["technical-description","exploit","evidence"]},{"url":"https://github.com/maziyarpanahi/openmed/releases/tag/v1.5.2","tags":["release-notes","advisory","web"]},{"url":"https://github.com/maziyarpanahi/openmed/pull/59","tags":["issue-tracking","patch","fix","web"]},{"url":"https://github.com/maziyarpanahi/openmed/commit/98724f65df98d7518b9006e6356740aa36c2f224","tags":["patch","fix","web"]},{"url":"https://www.vulncheck.com/advisories/openmed-remote-code-execution-via-pii-model-loading","tags":["third-party-advisory","advisory","web"]},{"url":"https://nvd.nist.gov/vuln/detail/CVE-2026-47117","tags":["advisory"]},{"url":"https://github.com/advisories/GHSA-m3v4-v5gx-7wf5","tags":["advisory"]},{"url":"https://github.com/CVEProject/cvelistV5/tree/main/cves/2026/47xxx/CVE-2026-47117.json","tags":["advisory"]},{"url":"https://github.com/maziyarpanahi/openmed","tags":["package"]},{"url":"github.com/maziyarpanahi/openmed","tags":["package"]},{"url":"https://pypi.org/project/openmed","tags":["package"]}],"sources":{"cve":[{"id":"CVE-2026-47117","title":"OpenMed \u003c 1.5.2 Remote Code Execution via PII Model Loading","description":"OpenMed before 1.5.2 contains a remote code execution vulnerability in the PII privacy-filter model loading path. The privacy-filter dispatcher used broad substring matching on the user-supplied model_name parameter, allowing a value such as attacker/foo-privacy-filter-bar to route through a path that loads Hugging Face models with trust_remote_code=True. An unauthenticated attacker can supply a malicious model repository containing custom Transformers code via auto_map in config.json or tokenizer_config.json, which is imported and executed with the privileges of the OpenMed service process.","status":"published","published":"2026-06-02T14:15:17.898Z","modified":"2026-11-08T14:04:08.243Z","cvss":[{"version":"4.0","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","level":"critical","assessor":"cna"},{"version":"3.1","score":9.8,"vector":"CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:H/A:H","level":"critical","assessor":"cna"}],"cwes":["CWE-94"],"updated_at":"2026-11-08T15:43:33.322657Z"}],"euvd":[{"id":"EUVD-2026-33942","description":"OpenMed before 1.5.2 contains a remote code execution vulnerability in the PII privacy-filter model loading path. The privacy-filter dispatcher used broad substring matching on the user-supplied model_name parameter, allowing a value such as attacker/foo-privacy-filter-bar to route through a path that loads Hugging Face models with trust_remote_code=True. An unauthenticated attacker can supply a malicious model repository containing custom Transformers code via auto_map in config.json or tokenizer_config.json, which is imported and executed with the privileges of the OpenMed service process.","status":"published","published":"2026-06-02T14:15:17Z","modified":"2026-11-08T14:04:08Z","cvss":[{"version":"4.0","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","level":"critical"}],"updated_at":"2026-11-08T15:13:08.77824Z"}],"ghsa":[{"id":"GHSA-m3v4-v5gx-7wf5","title":"OpenMed vulnerable to remote code injection through privacy-filter model loading path","description":"OpenMed before 1.5.2 contains a remote code execution vulnerability in the PII privacy-filter model loading path. The privacy-filter dispatcher used broad substring matching on the user-supplied `model_name` parameter, allowing a value such as `attacker/foo-privacy-filter-bar` to route through a path that loads Hugging Face models with `trust_remote_code=True`. An unauthenticated attacker can supply a malicious model repository containing custom Transformers code via auto_map in `config.json` or `tokenizer_config.json`, which is imported and executed with the privileges of the OpenMed service process.","status":"published","published":"2026-06-02T18:31:33Z","modified":"2026-06-30T21:42:34Z","cvss":[{"version":"4.0","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","level":"critical"},{"version":"3.1","score":9.8,"vector":"CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:H/A:H","level":"critical"}],"cwes":["CWE-94"],"updated_at":"2026-11-02T14:40:35.701766Z"}],"nvd":[{"id":"CVE-2026-47117","description":"OpenMed before 1.5.2 contains a remote code execution vulnerability in the PII privacy-filter model loading path. The privacy-filter dispatcher used broad substring matching on the user-supplied model_name parameter, allowing a value such as attacker/foo-privacy-filter-bar to route through a path that loads Hugging Face models with trust_remote_code=True. An unauthenticated attacker can supply a malicious model repository containing custom Transformers code via auto_map in config.json or tokenizer_config.json, which is imported and executed with the privileges of the OpenMed service process.","status":"published","published":"2026-06-02T16:16:43.98Z","modified":"2026-10-08T16:17:17.687Z","cvss":[{"version":"4.0","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/E:X/CR:X/IR:X/AR:X/MAV:X/MAC:X/MAT:X/MPR:X/MUI:X/MVC:X/MVI:X/MVA:X/MSC:X/MSI:X/MSA:X/S:X/AU:X/R:X/V:X/RE:X/U:X","level":"critical","assessor":"secondary"},{"version":"3.1","score":9.8,"vector":"CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:H/A:H","level":"critical","assessor":"secondary"}],"cwes":["CWE-94"],"updated_at":"2026-11-02T14:12:22.022466Z"}],"osv":[{"id":"CVE-2026-47117","title":"OpenMed \u003c 1.5.2 Remote Code Execution via PII Model Loading","description":"OpenMed before 1.5.2 contains a remote code execution vulnerability in the PII privacy-filter model loading path. The privacy-filter dispatcher used broad substring matching on the user-supplied model_name parameter, allowing a value such as attacker/foo-privacy-filter-bar to route through a path that loads Hugging Face models with trust_remote_code=True. An unauthenticated attacker can supply a malicious model repository containing custom Transformers code via auto_map in config.json or tokenizer_config.json, which is imported and executed with the privileges of the OpenMed service process.","status":"published","published":"2026-06-02T14:15:17.898Z","modified":"2026-11-11T02:47:57.181904425Z","cvss":[{"version":"4.0","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","level":"critical"}],"cwes":["CWE-94"],"updated_at":"2026-11-09T03:15:52.404586Z"},{"id":"GHSA-m3v4-v5gx-7wf5","title":"OpenMed vulnerable to remote code injection through privacy-filter model loading path","description":"OpenMed before 1.5.2 contains a remote code execution vulnerability in the PII privacy-filter model loading path. The privacy-filter dispatcher used broad substring matching on the user-supplied `model_name` parameter, allowing a value such as `attacker/foo-privacy-filter-bar` to route through a path that loads Hugging Face models with `trust_remote_code=True`. An unauthenticated attacker can supply a malicious model repository containing custom Transformers code via auto_map in `config.json` or `tokenizer_config.json`, which is imported and executed with the privileges of the OpenMed service process.","status":"published","published":"2026-06-02T18:31:33Z","modified":"2026-07-13T16:43:30.876937992Z","cvss":[{"version":"3.1","score":9.8,"vector":"CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:H/A:H","level":"critical"},{"version":"4.0","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","level":"critical"}],"cwes":["CWE-94"],"updated_at":"2026-11-02T15:46:51.10002Z"},{"id":"PYSEC-2026-2852","title":"OpenMed vulnerable to remote code injection through privacy-filter model loading path","description":"OpenMed before 1.5.2 contains a remote code execution vulnerability in the PII privacy-filter model loading path. The privacy-filter dispatcher used broad substring matching on the user-supplied `model_name` parameter, allowing a value such as `attacker/foo-privacy-filter-bar` to route through a path that loads Hugging Face models with `trust_remote_code=True`. An unauthenticated attacker can supply a malicious model repository containing custom Transformers code via auto_map in `config.json` or `tokenizer_config.json`, which is imported and executed with the privileges of the OpenMed service process.","status":"published","published":"2026-07-13T15:46:14.064755Z","modified":"2026-07-13T16:32:33.806674335Z","cvss":[{"version":"3.1","score":9.8,"vector":"CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:H/A:H","level":"critical"},{"version":"4.0","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","level":"critical"}],"updated_at":"2026-11-02T15:52:56.429374Z"}]}}