{"id":"CVE-2026-119328","aliases":["EUVD-2026-114375","GHSA-72rm-x7xr-v679"],"status":"published","title":"Immich: Unauthenticated filesystem path injection (arbitrary directory write + recursive delete) in the machine-learning service","description":"Immich is a high-performance self-hosted photo and video management solution. Prior to 3.3.0, the unauthenticated machine-learning /predict endpoint in machine-learning/immich_ml/main.py accepted an options object and passed its values into model construction in machine-learning/immich_ml/models/base.py without restricting security-sensitive parameters. An attacker with network access to the service could set cache_dir to an arbitrary writable path, causing model downloads to write there and causing clear_cache() to pass that path to shutil.rmtree() after selected model-loading failures. Depending on filesystem permissions and the selected path, this could modify files in mounted data locations or recursively delete an attacker-selected directory. The model-name allowlist prevents this issue from becoming arbitrary-model deserialization or code execution. This issue is fixed in version 3.3.0.","severity":{"version":"3.1","score":6.5,"vector":"CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:L/A:L","level":"medium","assessor":"cna","from":"cve"},"epss":{"day":"2026-11-11","score":0.00436,"percentile":0.38337,"model":"v2026.06.15"},"facts":{"in_kev":false,"known_exploited":false,"public_exploit":false,"fix_available":false},"cwes":["CWE-22","CWE-73"],"published":"2026-10-30T19:50:02Z","modified":"2026-11-06T17:38:35Z","products":[{"vendor":"immich-app","product":"immich"}],"references":[{"url":"https://github.com/immich-app/immich/security/advisories/GHSA-72rm-x7xr-v679","tags":["x_refsource_CONFIRM","advisory"]},{"url":"https://github.com/immich-app/immich/pull/31856","tags":["x_refsource_MISC","fix"]},{"url":"https://github.com/immich-app/immich/commit/ab159b81182f26a8943f5dd66b719e8cdaad4626","tags":["x_refsource_MISC","fix"]},{"url":"https://github.com/immich-app/immich/releases/tag/v3.3.0","tags":["x_refsource_MISC","web"]},{"url":"https://github.com/CVEProject/cvelistV5/tree/main/cves/2026/119xxx/CVE-2026-119328.json","tags":["advisory"]},{"url":"https://nvd.nist.gov/vuln/detail/CVE-2026-119328","tags":["advisory"]}],"sources":{"cve":[{"id":"CVE-2026-119328","title":"Immich: Unauthenticated filesystem path injection (arbitrary directory write + recursive delete) in the machine-learning service","description":"Immich is a high-performance self-hosted photo and video management solution. Prior to 3.3.0, the unauthenticated machine-learning /predict endpoint in machine-learning/immich_ml/main.py accepted an options object and passed its values into model construction in machine-learning/immich_ml/models/base.py without restricting security-sensitive parameters. An attacker with network access to the service could set cache_dir to an arbitrary writable path, causing model downloads to write there and causing clear_cache() to pass that path to shutil.rmtree() after selected model-loading failures. Depending on filesystem permissions and the selected path, this could modify files in mounted data locations or recursively delete an attacker-selected directory. The model-name allowlist prevents this issue from becoming arbitrary-model deserialization or code execution. This issue is fixed in version 3.3.0.","status":"published","published":"2026-10-30T19:50:02.46Z","modified":"2026-11-06T17:38:35.404Z","cvss":[{"version":"3.1","score":6.5,"vector":"CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:L/A:L","level":"medium","assessor":"cna"}],"cwes":["CWE-22","CWE-73"],"updated_at":"2026-11-06T18:59:07.714367Z"}],"euvd":[{"id":"EUVD-2026-114375","description":"Immich is a high-performance self-hosted photo and video management solution. Prior to 3.3.0, the unauthenticated machine-learning /predict endpoint in machine-learning/immich_ml/main.py accepted an options object and passed its values into model construction in machine-learning/immich_ml/models/base.py without restricting security-sensitive parameters. An attacker with network access to the service could set cache_dir to an arbitrary writable path, causing model downloads to write there and causing clear_cache() to pass that path to shutil.rmtree() after selected model-loading failures. Depending on filesystem permissions and the selected path, this could modify files in mounted data locations or recursively delete an attacker-selected directory. The model-name allowlist prevents this issue from becoming arbitrary-model deserialization or code execution. This issue is fixed in version 3.3.0.","status":"published","published":"2026-10-30T19:50:02Z","modified":"2026-11-06T17:38:35Z","cvss":[{"version":"3.1","score":6.5,"vector":"CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:L/A:L","level":"medium"}],"updated_at":"2026-11-06T19:05:53.354277Z"}],"nvd":[{"id":"CVE-2026-119328","description":"Immich is a high-performance self-hosted photo and video management solution. Prior to 3.3.0, the unauthenticated machine-learning /predict endpoint in machine-learning/immich_ml/main.py accepted an options object and passed its values into model construction in machine-learning/immich_ml/models/base.py without restricting security-sensitive parameters. An attacker with network access to the service could set cache_dir to an arbitrary writable path, causing model downloads to write there and causing clear_cache() to pass that path to shutil.rmtree() after selected model-loading failures. Depending on filesystem permissions and the selected path, this could modify files in mounted data locations or recursively delete an attacker-selected directory. The model-name allowlist prevents this issue from becoming arbitrary-model deserialization or code execution. This issue is fixed in version 3.3.0.","status":"published","published":"2026-10-30T20:17:06.627Z","modified":"2026-11-06T18:17:57.92Z","cvss":[{"version":"3.1","score":6.5,"vector":"CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:L/A:L","level":"medium","assessor":"secondary"}],"cwes":["CWE-22","CWE-73"],"updated_at":"2026-11-06T20:01:16.0125Z"}],"osv":[{"id":"CVE-2026-119328","title":"Immich: Unauthenticated filesystem path injection (arbitrary directory write + recursive delete) in the machine-learning service","description":"Immich is a high-performance self-hosted photo and video management solution. Prior to 3.3.0, the unauthenticated machine-learning /predict endpoint in machine-learning/immich_ml/main.py accepted an options object and passed its values into model construction in machine-learning/immich_ml/models/base.py without restricting security-sensitive parameters. An attacker with network access to the service could set cache_dir to an arbitrary writable path, causing model downloads to write there and causing clear_cache() to pass that path to shutil.rmtree() after selected model-loading failures. Depending on filesystem permissions and the selected path, this could modify files in mounted data locations or recursively delete an attacker-selected directory. The model-name allowlist prevents this issue from becoming arbitrary-model deserialization or code execution. This issue is fixed in version 3.3.0.","status":"published","published":"2026-10-30T19:50:02.46Z","modified":"2026-11-08T02:30:25.963652513Z","cvss":[{"version":"3.1","score":6.5,"vector":"CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:L/A:L","level":"medium"}],"cwes":["CWE-22","CWE-73"],"updated_at":"2026-11-07T03:08:01.41647Z"}]}}