RunSybil finds CPython flaw enabling sandbox escapes

Security researchers at RunSybil discovered a memory-safety flaw in CPython that can be exploited by malicious AI model templates to escape code sandboxes and execute arbitrary code in inference servers. The bug abuses CPython's handling of generic types to retrieve references to protected Python objects. While the Python Security Response Team acknowledged the underlying bug, it disputed the vulnerability classification, stating that sandboxing untrusted Python code is outside CPython's security threat model. The fix has been merged and backported as a regular bug fix rather than a security fix. Open article →