Back Redpacketsecurity CVE Alert: CVE-2026-120833 – NIST
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.
**Risk verdict:** High risk where untrusted data files are loaded; exploitation requires a user to open the file, but successful execution can fully compromise that user’s process.
**Why this matters:** An attacker could use a booby-trapped data file to steal accessible research or credentials, alter results, or disrupt analysis. No KEV, SSVC, PoC, or EPSS information is supplied, so active exploitation and prevalence cannot be established.
**Most likely attack path:** An attacker gets a crafted file onto a researcher’s workstation or shared analysis environment, then persuades a user to load it; no prior privileges or complex conditions are indicated. Impact is confined to the affected process by the supplied scope assessment, though access to its files, credentials, or network connections could enable further compromise.
**Who is most exposed:** Research and engineering teams using Python-based laboratory data workflows are most exposed, particularly where files arrive from external collaborators or are processed from shared repositories and notebooks.
Review application and endpoint logs for data-loading activity involving pickle-format files.
Hunt for unexpected child processes, shell activity, or outbound connections from analysis environments after file loads.
Check shared directories, notebooks, and recent downloads for untrusted data archives.
Mitigation and prioritisation
Upgrade promptly to a fixed release; prioritise systems that load external or shared data.
Until patched, block or quarantine untrusted pickle-format files and use safer, explicitly validated formats.
Restrict analysis environments’ access to secrets, sensitive files, and outbound network connections.
Coordinate upgrades with researchers, then validate representative workflows and monitor for regressions.
A considerable amount of time and effort goes into maintaining this website, creating backend automation and creating new features and content for you to make actionable intelligence decisions. Everyone that supports the site helps enable new functionality.
If you like the site, please support us on Patreon or Buy Me A Coffee using the buttons below.
The full story
This article is one source in a clustered incident — the cluster page carries the summary, timeline and every other outlet covering it.
