This vulnerability in picklescan prior to version 0.0.28 involves a failure to detect malicious pickle files that exploit the torch._dynamo.guards.GuardBuilder.get function used in reduce methods. Attackers can create specially crafted pickle files that evade detection by picklescan and execute arbitrary code when these files are deserialized. The issue relates to unsafe deserialization practices and is tracked as CVE-2025-71353. There is no patch or remediation information provided, and no known exploits in the wild have been reported.

Successful exploitation allows attackers to execute arbitrary commands on the system loading the malicious pickle file, potentially leading to full compromise of the affected environment. This poses a high security risk due to the ability to run unauthorized code remotely or locally via crafted pickle files.

Patch status is not yet confirmed — check the vendor advisory for current remediation guidance. Until an official fix is available, avoid loading pickle files from untrusted or unauthenticated sources. Monitor vendor channels for updates regarding patches or official mitigations.