What is pickling and what are its risks?

Question

Accepted Answer

**Pickling** serializes almost any Python object graph into a byte stream (`pickle.dumps`) and reconstructs it later (`pickle.loads`) — preserving types, nesting and references. It's Python's native persistence/IPC format and is what `multiprocessing` uses to pass objects between processes.

The critical risk: **`pickle.loads` on untrusted data can execute arbitrary code.** Unpickling can invoke an object's `__reduce__`, which may call any callable — so a malicious pickle can run `os.system(...)` the moment you load it. **Never unpickle data from an untrusted or unauthenticated source.**

Other limitations: it's **Python-specific** (not interoperable with other languages), **not human-readable**, can **break across Python/library versions**, and can't pickle some objects (open files, sockets, lambdas, local functions).

Safer alternatives: **JSON** for interoperable, human-readable, untrusted data (limited to basic types); MessagePack/Protocol Buffers for compact cross-language formats

What is pickling and what are its risks?

Understand the problem

Attempt it yourself

Stuck? AI Nudge Available

Study the solution

The solution is waiting

Run the code

Join the discussion

Discussion (0)