Skip to content

SecurityAmazon

Workaround available

Amazon: Cairn AI agents found and exploited a MONAI Label zero-day (CVE-2026-133968)

MONAI Label at a glance

Amazon Threat Intelligence disclosed on November 20, 2026 that a Chinese-nexus actor's Cairn AI agents autonomously found and exploited a zero-day in MONAI Label (CVE-2026-133968, CVSS 9.8), reaching root on a university server; no patched release exists.

Ask your AI about this story

Ask what changed in MONAI Label and Cairn and what to check before using it in your business. Identify affected versions, the vendor's mitigation or fix, and how to verify that it is in place.

Sources checked Primary source (opens in a new tab)

MONAI Label: status

Workaround availableCritical
Affected
  • MONAI Label 0.8.5 and earlier (pip package monailabel): a caller-controlled image parameter in PUT /datastore/image, and the same pattern in PUT /datastore/label, allows arbitrary file write and remote code execution
  • Deployments with authentication disabled, the default (MONAI_LABEL_AUTH_ENABLE=False), where any network client can exploit the flaw; with authentication enabled, the ANNOTATOR role is sufficient
  • Two East Asian university networks targeted September 8-10, 2026; one server was taken over with root access and the same flaw was inferred in the second network
Last checked
Discovered by
Autonomous Cairn AI agents run by a suspected Chinese-nexus actor; disclosed by Amazon Threat Intelligence · Found by an AI system

What changed with MONAI Label

Amazon Threat Intelligence disclosed on November 20, 2026 that autonomous AI agents running on Cairn, an open-source penetration-testing framework, found and exploited a zero-day in MONAI Label, an open-source medical-imaging AI application. Amazon found the actor's unauthenticated Cairn server on September 10, 2026 and reconstructed a campaign of about 48 hours from September 8, 2026 against universities in East Asia outside mainland China: given the goal 'obtain shell access,' the agents mapped MONAI Label's API, tested its upload endpoint, and went from mapping to arbitrary file writes in about 11 minutes, later writing a Python reverse shell into sitecustomize.py and triggering a training endpoint to run it as root.

The flaw, CVE-2026-133968, is a path traversal in MONAI Label's PUT /datastore/image and PUT /datastore/label endpoints: the National Vulnerability Database entry published on November 20, 2026 scores it 9.8 of 10 (critical, CVSS v3.1) and lists versions 0.8.5 and earlier as affected, with authentication disabled by default. The agents succeeded only there: one operation ran 148 sequential actions over about 16 hours against a Redis server without success, another tested more than 30,000 credential combinations against a university GPU server, and the actor later reused the zero-day through a proxy tunnel into a second university network. Project MONAI merged the code fix (pull request #1915) on October 19, 2026 and published advisory GHSA-qc22-rmgc-vg94 on November 18, 2026, but no patched release exists as of November 20, 2026.

How to fix it: MONAI Label

  1. 1. Workaround

    Enable authentication (MONAI_LABEL_AUTH_ENABLE=True) and do not grant the ANNOTATOR role to untrusted users.

    github.com (opens in a new tab)
  2. 2. Workaround

    Do not expose the MONAI Label server to untrusted networks until a patched release is installed.

    github.com (opens in a new tab)
  3. 3. Configuration

    Run the server as an unprivileged user with write access limited to the datastore directory.

    github.com (opens in a new tab)
  4. 4. Update

    Apply the code fix merged in pull request #1915 when the next MONAI Label release ships; no patched release exists as of November 20, 2026.

    github.com (opens in a new tab)
  5. 5. Monitor

    Prioritize high-confidence detections that trigger an automated response, with humans overseeing outcomes rather than reviewing every individual action (Amazon Threat Intelligence guidance).

    aws.amazon.com (opens in a new tab)

Advisories

MONAI Label in numbers

Values are copied from the linked source on the date shown. They are not IA Compa measurements.

CVE-2026-133968 CVSS v3.1 base score (NVD)as of · Source for CVE-2026-133968 CVSS v3.1 base score (NVD): nvd.nist.gov (opens in a new tab)
9.8
Minutes from API mapping to arbitrary file write (Amazon) · Cairnas of · Source for Minutes from API mapping to arbitrary file write (Amazon): aws.amazon.com (opens in a new tab)
11
Sequential agent actions in the failed Redis operation (Amazon) · Cairnas of · Source for Sequential agent actions in the failed Redis operation (Amazon): aws.amazon.com (opens in a new tab)
148
Credential combinations tested against a university GPU server (Amazon) · Cairnas of · Source for Credential combinations tested against a university GPU server (Amazon): aws.amazon.com (opens in a new tab)
30000

Why MONAI Label matters for businesses

Why it matters: offensive agents now compress the time between a flaw existing and a server falling to minutes, while many businesses still patch on weekly or monthly cycles, so any company running AI servers, imaging tools or agent frameworks with default settings sits exposed in that gap. A business can use the same approach defensively: run agent-based penetration testing against its own systems before someone else does, switch on authentication and least-privilege access for every AI tool it hosts, and automate first-response actions so people review outcomes instead of triaging every alert by hand.

Get AI News by email

Verified AI stories, each with its official source, on Monday and Thursday mornings. Add breaking news to hear about the biggest ones as soon as we publish them.

Every story links its official source

Free, in English or Spanish

Unsubscribe with one click, any time

What to send

Sources for this story

  1. Official
  2. Official
  3. CVE-2026-133968 (opens in a new tab)

    National Vulnerability Database · nvd.nist.gov

    Official

Last verified: . IA Compa summarizes public announcements for business readers. We are not affiliated with the companies mentioned. Always confirm details, pricing, and availability with the official source.

  • #cairn
  • #monai-label
  • #zero-day
  • #cve-2026-133968
  • #path-traversal
  • #autonomous-ai
  • #threat-intelligence
  • #ai-security

Want this working in your business?

We turn new AI capabilities into agents, voice receptionists, and automations you can actually use.