GHSA-8r4j-24qv-fmq9: picklescan: RCE bypass enables ML supply chain attack

### Summary Using idlelib.calltip.Calltip.fetch_tip, 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 idlelib.calltip.Calltip.fetch_tip 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 ``` class EvilCalltipFetchTip: def __reduce__(self): from idlelib.calltip import Calltip # fetch_tip(expression) -> get_entity(expression) -> eval(expression) return Calltip().fetch_tip, ("__import__('os').system('whoami')",) ``` ### 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 attacker uploads a malicious PyTorch model to HuggingFace Hub or a public model registry. The file contains a pickle payload using idlelib.calltip.Calltip.fetch_tip in its __reduce__ method — a Python built-in that internally calls eval(), executing arbitrary OS commands. An ML engineer discovers the model, runs picklescan for safety validation, and receives a CLEAN result. Trusting the tool, the engineer loads the model locally or triggers it through a CI/CD pipeline. RCE fires: the attacker gains a shell in the engineer's environment or production inference infrastructure, with potential access to model weights, training datasets, API keys, and lateral movement into internal systems. The entire attack chain requires no zero-day and is replicable by any attacker who reads the public advisory.