GHSA-p9w7-82w4-7q8m: picklescan: detection bypass allows pickle RCE in ML pipelines

### Summary Using lib2to3.pgen2.pgen.ParserGenerator.make_label function, which is a built-in python library function to execute remote pickle file. ### Details The attack payload executes in the following steps: First, the attacker craft the payload by calling to lib2to3.pgen2.pgen.ParserGenerator.make_label function in reduce method Then when the victim after checking whether the pickle file is safe by using Picklescan library and this library doesn't dectect any dangerous functions, decide to pickle.load() this malicious pickle file, thus lead to remote code execution. ### PoC ``` from types import SimpleNamespace from lib2to3.pgen2.pgen import ParserGenerator class EvilLib2to3Pgen2ParserGeneratorMakeLabel: def __reduce__(self): c = SimpleNamespace(labels=[], keywords={}, tokens={}) label = '""+__import__(\'os\').system(\'whoami\')' return ParserGenerator.make_label, (None, c, label) ``` ### Impact Who is impacted? Any organization or individual relying on picklescan to detect malicious pickle files inside PyTorch models. What is the impact? Attackers can embed malicious code in pickle file that remains undetected but executes when the pickle file is loaded. Supply Chain Attack: Attackers can distribute infected pickle files across ML models, APIs, or saved Python objects. ### Corresponding https://github.com/FredericDT https://github.com/Qhaoduoyu

An adversary targets an ML team that uses picklescan to validate PyTorch models before loading. The attacker crafts a malicious .pkl file using the lib2to3.pgen2.pgen.ParserGenerator.make_label gadget — a built-in Python stdlib function that picklescan does not flag. The payload is uploaded to a public model repository (e.g., HuggingFace) or embedded in a poisoned model shared via a community forum. The victim's automated pipeline scans the file with picklescan (result: clean), proceeds to torch.load() or pickle.load(), and the payload executes — running arbitrary OS commands (e.g., establishing a reverse shell, exfiltrating API keys, or pivoting to the training cluster).