CVE-2024-5452: pytorch-lightning remote code execution via deepdiff Delta input

Summary

A remote code execution flaw exists in the lightning-ai/pytorch-lightning library, version 2.2.1, caused by improper handling of deserialized user input and mismanaged dunder attributes in the deepdiff library. An attacker can craft a serialized deepdiff.Delta that passes the deserializer whitelist and contains dunder attributes, which lets them reach other modules, classes and instances, enabling arbitrary attribute write and full RCE on self-hosted applications in their default configuration because the delta endpoint is enabled by default.

Mitigation

The source does not state a fix yet. Check the original advisory for updates.

Affected packages in the Exposure Registry

Matched by package name and ecosystem. Each entry shows whether the package delegates to a language model and how many tracked packages depend on it.

• lightningPyPILLM dependency since 2025-10-26 · 3 tracked dependents

Related items

• MediumCVE-2026-121936: easyVoice server-side request forgery through openaiBaseUrl in /api/v1/ttsSimilar attack · NVD/CVE Database
• MediumCVE-2026-118628: MLflow get-artifacts server-side request forgery information disclosureSimilar attack · NVD/CVE Database
• HighCVE-2026-118625: MLflow Tracking Server privilege escalation via get_endpointsSimilar attack · NVD/CVE Database
• InfoWhen trust becomes the payload: Inside the threats, etc quarterly reportSimilar attack · Elastic Security Labs
• MediumCritical Langflow flaws enable authentication bypass and RCESimilar attack · CSO Online