Which versions of PyTorch Lightning are affected by CVE-2026-44484?

PyTorch Lightning versions 2.6.2 and 2.6.3 contain malicious code injected at the supply chain level that harvests API keys, access tokens, and SSH credentials from affected systems.

How to fix or mitigate CVE-2026-44484?

Within 24 hours: audit all systems and CI/CD pipelines for PyTorch Lightning 2.6.2 or 2.6.3 installation; immediately revoke all exposed API keys, access tokens, SSH keys, and service account credentials; review access logs for unauthorized activity. Within 7 days: downgrade all affected instances to version 2.6.1 or earlier confirmed clean release; patch all compromised credentials across dependent systems (cloud providers, code repositories, deployment platforms). Within 30 days: implement dependency pinning and software composition analysis (SCA) tooling to detect supply chain anomalies; conduct forensic analysis of compromised systems for persistent access mechanisms.

PyTorch Lightning CVE-2026-44484

Q: What is the CVSS score of CVE-2026-44484?

CVE-2026-44484 has a CVSS 4.0 base score of 9.3 (Critical). CVSS vector: CVSS:4.0/AV:N/AC:L/AT:N/PR:N/UI:N/VC:H/VI:H/VA:H/SC:N/SI:N/SA:N/E:X/CR:X/IR:X/AR:X/MAV:X/MAC:X/MAT:X/MPR:X/MUI:X/MVC:X/MVI:X/MVA:X/MSC:X/MSI:X/MSA:X/S:X/AU:X/R:X/V:X/RE:X/U:X. EPSS exploitation probability: 0.0%.

EUVD-2026-30303 CRITICAL

Embedded Malicious Code (CWE-506)

2026-05-07 https://github.com/Lightning-AI/pytorch-lightning

GHSA-w37p-236h-pfx3

Information Disclosure

9.3

CVSS 4.0 · Vendor: https://github.com/Lightning-AI/pytorch-lightning

Severity by source

Vendor (https://github.com/Lightning-AI/pytorch-lightning) PRIMARY

9.3 CRITICAL

CVSS:4.0/AV:N/AC:L/AT:N/PR:N/UI:N/VC:H/VI:H/VA:H/SC:N/SI:N/SA:N/E:X/CR:X/IR:X/AR:X/MAV:X/MAC:X/MAT:X/MPR:X/MUI:X/MVC:X/MVI:X/MVA:X/MSC:X/MSI:X/MSA:X/S:X/AU:X/R:X/V:X/RE:X/U:X

Red Hat

9.8 HIGH

qualitative

Primary rating from Vendor (https://github.com/Lightning-AI/pytorch-lightning).

CVSS VectorVendor: https://github.com/Lightning-AI/pytorch-lightning

CVSS:4.0/AV:N/AC:L/AT:N/PR:N/UI:N/VC:H/VI:H/VA:H/SC:N/SI:N/SA:N/E:X/CR:X/IR:X/AR:X/MAV:X/MAC:X/MAT:X/MPR:X/MUI:X/MVC:X/MVI:X/MVA:X/MSC:X/MSI:X/MSA:X/S:X/AU:X/R:X/V:X/RE:X/U:X

Attack Vector

Network

Attack Complexity

Low

Privileges Required

None

User Interaction

None

Scope

Lifecycle Timeline

Analysis Updated

May 14, 2026 - 15:27 vuln.today

v2 (cvss_changed)

Re-analysis Queued

May 14, 2026 - 15:22 vuln.today

cvss_changed

CVSS changed

May 14, 2026 - 15:22 NVD

9.3 (CRITICAL)

Source Code Evidence Fetched

May 07, 2026 - 01:15 vuln.today

Analysis Generated

May 07, 2026 - 01:15 vuln.today

CVE Published

May 07, 2026 - 00:52 nvd

CRITICAL

DescriptionCVE.org

Security Advisory: Compromise of PyTorch Lightning PyPI Package Versions

Published: 2026-04-30 Last Updated: 2026-04-30

Lightning AI has identified a security incident affecting certain versions of a PyPI package.

What happened

Lightning AI has determined that one or more released versions of this package have been compromised and include malicious code.

The current investigation indicates that the affected versions have introduced functionality consistent with a credential harvesting mechanism. There is a continuing analysis the scope and behaviour of the code.

At this stage, the root cause of the compromise is still under investigation.

What versions are affected

Lightning AI is currently working to confirm the exact set of impacted versions.

The following versions are determined as affected, and developers should delete them from their systems:

2.6.2
2.6.3

Lightning AI will update this advisory if the versions impacted by this vulnerability change.

What you should do immediately

If developers have installed or are running any potentially affected versions on their application:

Assume the environment may be compromised
Immediately rotate all credentials and secrets that may have been exposed, including:
API keys
Access tokens
SSH keys
Service account credentials
Rebuild affected systems from a known clean state
Pin PyTorch Lightning to version 2.6.1
Review logs for any suspicious or unauthorised activity

Actions Lightning AI has taken

Quarantined malicious versions from PyPI
Recommended using version 2.6.1: https://github.com/Lightning-AI/pytorch-lightning/releases/tag/2.6.1
Revoked and rotated all internal credentials associated with our release process
Initiated a full investigation into the compromise

Ongoing investigation

Lightning AI is actively working to:

Identify the exact mechanism of compromise
Confirm the full set of affected versions
Determine the behaviour and impact of the malicious code
Assess any downstream impact to users

Lightning AI will provide updates as soon as more information becomes available.

Commitment to transparency

Lightning AI takes the security of users and the integrity of the software supply chain extremely seriously.

Lightning AI will continue to share timely and accurate updates as the investigation progresses.

Contact

If there are any questions or if there are concerns that a consuming project may be impacted, please send an email to:

security@lightning.ai

AnalysisAI

Supply chain compromise in PyTorch Lightning versions 2.6.2 and 2.6.3 delivers credential-harvesting malware through the official PyPI distribution channel. Lightning AI confirmed malicious code was injected into these specific releases, targeting API keys, access tokens, SSH keys, and service account credentials. The compromised versions were quarantined from PyPI and users are directed to downgrade to known-clean version 2.6.1. The CVSS 9.3 score reflects network-accessible exploitation requiring no authentication or user interaction, though exploitation is limited to systems that specifically installed the two poisoned versions.

Technical ContextAI

This is a software supply chain attack (CWE-506: Embedded Malicious Code) targeting the Python Package Index (PyPI) distribution mechanism for pytorch-lightning. The malicious code was inserted directly into official package releases 2.6.2 and 2.6.3, making this a compromise of the software build/release pipeline rather than a traditional code vulnerability. The credential harvesting mechanism embedded in these versions operates at the Python package level, executing during installation or import of the compromised library. This attack vector exploits the trust relationship between developers and the PyPI ecosystem, similar to previous supply chain incidents affecting other Python packages. The CPE identifier pkg:pip/pytorch-lightning confirms the Python pip package manager distribution channel was the attack vector.

RemediationAI

Immediately audit all systems for pytorch-lightning installations using 'pip list | grep pytorch-lightning' or equivalent package manager commands. If versions 2.6.2 or 2.6.3 are detected, treat the entire environment as compromised: rotate ALL credentials potentially accessible to the runtime environment including API keys, access tokens, SSH keys, database passwords, cloud service credentials, and certificate private keys. Rebuild affected systems from known-clean state or at minimum uninstall the compromised package and install version 2.6.1 using 'pip install pytorch-lightning2.6.1'. Pin the version in requirements.txt or pyproject.toml to prevent automatic upgrades: 'pytorch-lightning2.6.1'. Review application logs, CloudTrail/Azure Activity logs, and authentication logs for unauthorized access patterns during the exposure window. Monitor the vendor advisory (https://github.com/Lightning-AI/pytorch-lightning/security/advisories/GHSA-w37p-236h-pfx3) for updates as investigation may reveal additional compromised versions or specific indicators of compromise. NO technical workaround exists for already-installed malicious code beyond complete removal and credential rotation.