MONAI 1.2.0 before 1.6.1 contains an insecure deserialization vulnerability in nnUNetV2Runner that unpickles the postprocessingfile path read from inferenceinformation.json without confining it to the results directory. Attackers who can modify inferenceinformation.json can point it at a malicious pickle elsewhere on the filesystem to execute code when predictensemblepostprocessing runs.

MONAI before 1.5.2 contains a deserialization of untrusted data vulnerability in the algofrompickle function in monai/auto3dseg/utils.py. The function reads a .pkl file and passes its contents to pickle.loads without validating the data source or content. If an application invokes algofrompickle on an attacker-supplied pickle file, an object defining reduce is executed during deserialization, resulting in arbitrary code execution in the context of the application.

MONAI through 1.6.0 contains an eval injection vulnerability in getfakespatialshape() in monai/bundle/scripts.py. The function validates shape expressions with a helper that walks the AST and only collects ast.Name nodes, rejecting any name other than 'p' or 'n', before passing the string to eval(). Expressions built solely from constants and attribute, subscript, or call nodes (for example "(1).class.bases[0].subclasses()" or "int.class.init.globals") contain no ast.Name nodes and therefore bypass the allowlist. Because the shape value originates from bundle metadata consumed by getrealinputdata and verifynetinout (reachable through the bundle 'verifynetinout' CLI flow), an attacker who can influence a bundle's metadata can escape the eval sandbox via object introspection chains and achieve code execution in this non-default flow.