ultralytics 8.2.95

Direct Vulnerabilities

Known vulnerabilities in the ultralytics package. This does not include vulnerabilities belonging to this package’s dependencies.

Fix vulnerabilities automatically

Snyk's AI Trust Platform automatically finds the best upgrade path and integrates with your development workflows. Secure your code at zero cost.

Fix for free

Vulnerability	Vulnerable Version
H Arbitrary Code Injection ultralytics is an Ultralytics YOLOv8 for SOTA object detection, multi-object tracking, instance segmentation, pose estimation and image classification. Affected versions of this package are vulnerable to Arbitrary Code Injection via the `check_det_dataset` function in `ultralytics/data/utils.py`. An attacker can execute arbitrary Python code in the current user’s context by supplying a malicious dataset YAML whose `download` value is executed when the requested validation split is missing and autodownload is enabled; exploitation requires the user to open the malicious file or visit a malicious page. How to fix Arbitrary Code Injection? There is no fixed version for `ultralytics`.	[0,)
H Eval Injection ultralytics is an Ultralytics YOLOv8 for SOTA object detection, multi-object tracking, instance segmentation, pose estimation and image classification. Affected versions of this package are vulnerable to Eval Injection via the `parse_model` function in `ultralytics/nn/tasks.py`. An attacker can execute arbitrary Python code in the current user’s context by getting them to open or load a malicious model file whose `activation` value is evaluated without validation; exploitation requires user interaction and occurs in the default, non-restricted loading path. How to fix Eval Injection? There is no fixed version for `ultralytics`.	[0,)
H Deserialization of Untrusted Data ultralytics is an Ultralytics YOLOv8 for SOTA object detection, multi-object tracking, instance segmentation, pose estimation and image classification. Affected versions of this package are vulnerable to Deserialization of Untrusted Data via the `load_dataset_cache_file` function in `ultralytics/data/utils.py`. An attacker can execute arbitrary code in the current user’s process by getting them to open a dataset containing a malicious `.cache` file; deserialization occurs during dataset initialization before cache version and hash checks. How to fix Deserialization of Untrusted Data? There is no fixed version for `ultralytics`.	[0,)
H Arbitrary Code Injection ultralytics is an Ultralytics YOLOv8 for SOTA object detection, multi-object tracking, instance segmentation, pose estimation and image classification. Affected versions of this package are vulnerable to Arbitrary Code Injection via the `parse_model` function in `ultralytics/nn/tasks.py`. An attacker can execute arbitrary Python code in the current user’s process by supplying a malicious model YAML whose `activation` value is evaluated when the target opens or loads the file. How to fix Arbitrary Code Injection? There is no fixed version for `ultralytics`.	[0,)
H Directory Traversal ultralytics is an Ultralytics YOLOv8 for SOTA object detection, multi-object tracking, instance segmentation, pose estimation and image classification. Affected versions of this package are vulnerable to Directory Traversal via the `safe_download` function in `ultralytics/utils/downloads.py`. An attacker can execute arbitrary code in the current process by getting a target to visit a malicious page or open a malicious file that triggers a Google Drive download whose response supplies a traversal filename in its `Content-Disposition` header; `safe_download` uses that unvalidated filename to construct the output path and write outside the intended destination. How to fix Directory Traversal? There is no fixed version for `ultralytics`.	[0,)
H Deserialization of Untrusted Data ultralytics is an Ultralytics YOLOv8 for SOTA object detection, multi-object tracking, instance segmentation, pose estimation and image classification. Affected versions of this package are vulnerable to Deserialization of Untrusted Data via the `torch_load` function in `ultralytics/utils/patches.py`. An attacker can execute arbitrary code in the current user’s context by supplying a malicious checkpoint that is deserialized without restricted loading. Exploitation requires the user to open the malicious file through Ultralytics. How to fix Deserialization of Untrusted Data? There is no fixed version for `ultralytics`.	[0,)
H Command Injection ultralytics is an Ultralytics YOLOv8 for SOTA object detection, multi-object tracking, instance segmentation, pose estimation and image classification. Affected versions of this package are vulnerable to Command Injection via the `torch_safe_load` function in `ultralytics/nn/tasks.py`. An attacker can execute arbitrary code as the current user by tricking a victim into opening or loading a malicious checkpoint that supplies a crafted missing-module name; `torch_safe_load` passes that name to dependency checking, which may install a package. How to fix Command Injection? There is no fixed version for `ultralytics`.	[0,)
H Arbitrary Code Injection ultralytics is an Ultralytics YOLOv8 for SOTA object detection, multi-object tracking, instance segmentation, pose estimation and image classification. Affected versions of this package are vulnerable to Arbitrary Code Injection via unsafe use of `eval()` on attacker-controllable strings. The `cfg.smart_value` helper, the `string-handling` branch in `utils.checks.check_imgsz`, and the initializer in `utils/triton.py` evaluate arbitrary text with `eval()` (for example, when parsing imgsz or Triton metadata) instead of using a safe parser like `ast.literal_eval()`. An attacker who can influence these inputs can inject and execute arbitrary Python expressions in the context of the running process. How to fix Arbitrary Code Injection? Upgrade `ultralytics` to version 8.3.226 or higher.	[,8.3.226)

Vulnerability

Vulnerable Version

Arbitrary Code Injection