---
id: "CVE-2026-120833"
kind: "vulnerability"
aliases: ["CVE-2026-120833"]
title: "labbench: crafted data files run code when loaded with read()"
severity: "high"
cvss: 8.5
epss: null
exploited: false
status: "open"
category: "package"
products: []
ecosystems: []
tags: ["data-pipelines", "labbench", "package", "python"]
not_affected_if: ["labbench-0.46.1-or-later", "no-untrusted-data-files"]
published_at: "2026-11-03T01:16:52+00:00"
first_seen_at: "2026-11-03T01:47:23+00:00"
updated_at: "2026-11-03T01:48:23+00:00"
sources: [{"source": "nvd", "url": "https://nvd.nist.gov/vuln/detail/CVE-2026-120833", "first_seen_at": "2026-11-03T01:47:23+00:00"}]
---

# labbench: crafted data files run code when loaded with read()

labbench before 0.46.1 passes any file ending in .p or .pickle to pandas.read_pickle based on the extension alone, in its read() and read_relational() functions. An attacker who supplies a crafted data file or relational data archive can run arbitrary code when a user loads it.

_Summary, check and fix written by a model from the sources; labelled as such._

## How to check

- Check the installed labbench version with `pip show labbench` and confirm whether it is below `0.46.1`.
- Search your code and notebooks for calls to `labbench` `read()` or `read_relational()` on files supplied by others.

## Fix

- labbench: move to `0.46.1` or later.

## Timeline

- 2026-11-03T01:47:23+00:00 nvd https://nvd.nist.gov/vuln/detail/CVE-2026-120833

## Documents

- [advisory] Third-party advisory: www.vulncheck.com https://www.vulncheck.com/advisories/labbench-before-0.46.1-deserialization-of-untrusted-data-via-read-pickle-auto-dispatch
- [fix] Patch: github.com https://github.com/usnistgov/labbench/commit/ac9b9b18cba67cb8e85eb563b6bb7dc1092b4c54

## From the sources

labbench before 0.46.1 contains a deserialization of untrusted data vulnerability in read() and read_relational() that passes any file with a .p or .pickle extension to pandas.read_pickle based on the extension alone. Attackers who supply a crafted data file or relational data archive can execute arbitrary code when a user loads it with these functions.

---

_Free to read at a human pace. An API with keys and an MCP server are in preparation; early access: hello@infrastack.ai_