CVE-2021-42950: n/a RCE — UNKNOWN

CISO Take

Zepl Notebooks, a SaaS data-science notebook platform, contained a remote code execution flaw where any self-registered trial user could create a notebook, plant specially crafted code, and execute arbitrary commands beyond the intended notebook sandbox. There's no CVSS score, no CISA KEV listing, no public exploit, and no Nuclei template, and EPSS sits at 0.015 — objectively low near-term exploitation probability despite ranking in the top 29th percentile of scored CVEs. The real concern for CISOs isn't mass exploitation but blast radius in a multi-tenant SaaS notebook: a single malicious free-trial account could pivot from its own workspace into host infrastructure or adjacent tenants' data and model artifacts. All versions before October 25, 2021 are affected; since Zepl is a hosted SaaS product, confirm with the vendor that the fix has been fully rolled out and audit any historical trial or free-tier accounts for anomalous notebook execution activity around that timeframe. Treat this as a low-urgency legacy finding to close out in vendor risk records rather than an active threat requiring emergency action.

Sources: NVD EPSS CISA KEV ATLAS

What is the risk?

Overall risk is LOW given the absence of a CVSS score, no CISA KEV listing, no observed in-the-wild exploitation, no public exploit code, and no Nuclei scanning template. EPSS of 0.015 confirms low probability of opportunistic exploitation. However, the technical severity of the underlying flaw is meaningful — unsandboxed or weakly-sandboxed remote code execution in a multi-tenant SaaS platform is a high-impact bug class even without active exploitation, since it only requires a free/trial-tier authenticated account (a very low barrier to entry) rather than any special privilege. This is a case where likelihood is low but impact-if-exploited is disproportionately high due to the multi-tenant SaaS deployment model.

How does the attack unfold?

Account Registration

Attacker self-registers a free/trial Zepl account and creates an organization, requiring no special privileges.

AML.T0021

Malicious Notebook Creation

Attacker creates a new notebook and authors specially crafted code in a supported language.

AML.T0050

Sandbox Escape

Executing the crafted notebook code breaks out of the intended execution isolation to run on the underlying host or shared infrastructure.

AML.T0105

Impact

Attacker achieves remote code execution on Zepl's infrastructure, risking exposure of other tenants' notebooks, data, and model artifacts.

Account Registration

Attacker self-registers a free/trial Zepl account and creates an organization, requiring no special privileges.

AML.T0021

Malicious Notebook Creation

Attacker creates a new notebook and authors specially crafted code in a supported language.

AML.T0050

Sandbox Escape

Executing the crafted notebook code breaks out of the intended execution isolation to run on the underlying host or shared infrastructure.

AML.T0105

Impact

Attacker achieves remote code execution on Zepl's infrastructure, risking exposure of other tenants' notebooks, data, and model artifacts.

What systems are affected?

Package	Ecosystem	Vulnerable Range	Patched
n/a	—	—	No patch

Do you use n/a? You're affected.

How severe is it?

CVSS 3.1

N/A

EPSS

1.6%

chance of exploitation in 30 days

Higher than 74% of all CVEs

Source: EPSS v3 — FIRST.org

Exploitation Status

No known exploitation

Sophistication

Moderate

What should I do?

1 step

1) Confirm with Zepl that all production environments were remediated after October 25, 2021 (the vendor-stated fix date) — since this is SaaS, no customer-side patching action is typically required, but vendor risk/procurement records should note the closure date. 2) For any organization still running a self-hosted or legacy Zepl deployment, upgrade immediately to a version released after October 25, 2021. 3) Audit historical trial/free-tier signups and notebook creation logs from before the fix date for anomalous process spawns, container escapes, or unexpected outbound network activity. 4) As a general control for any notebook-based SaaS platform, enforce strict sandbox/container isolation (gVisor, Firecracker, or equivalent), monitor kernel processes for host-breakout indicators, and limit blast radius by hard multi-tenant network segmentation.

How is it classified?

Code Execution Framework AML.T0050 - Command and Scripting Interpreter AML.T0105 - Escape to Host

Which compliance frameworks are affected?

This CVE is relevant to:

EU AI Act

Article 15 - Accuracy, robustness and cybersecurity

NIST AI RMF

GOVERN 6.1 - Third-party AI system risk management

Frequently Asked Questions

What is CVE-2021-42950?

Zepl Notebooks, a SaaS data-science notebook platform, contained a remote code execution flaw where any self-registered trial user could create a notebook, plant specially crafted code, and execute arbitrary commands beyond the intended notebook sandbox. There's no CVSS score, no CISA KEV listing, no public exploit, and no Nuclei template, and EPSS sits at 0.015 — objectively low near-term exploitation probability despite ranking in the top 29th percentile of scored CVEs. The real concern for CISOs isn't mass exploitation but blast radius in a multi-tenant SaaS notebook: a single malicious free-trial account could pivot from its own workspace into host infrastructure or adjacent tenants' data and model artifacts. All versions before October 25, 2021 are affected; since Zepl is a hosted SaaS product, confirm with the vendor that the fix has been fully rolled out and audit any historical trial or free-tier accounts for anomalous notebook execution activity around that timeframe. Treat this as a low-urgency legacy finding to close out in vendor risk records rather than an active threat requiring emergency action.

Is CVE-2021-42950 actively exploited?

No confirmed active exploitation of CVE-2021-42950 has been reported, but organizations should still patch proactively.

How to fix CVE-2021-42950?

1) Confirm with Zepl that all production environments were remediated after October 25, 2021 (the vendor-stated fix date) — since this is SaaS, no customer-side patching action is typically required, but vendor risk/procurement records should note the closure date. 2) For any organization still running a self-hosted or legacy Zepl deployment, upgrade immediately to a version released after October 25, 2021. 3) Audit historical trial/free-tier signups and notebook creation logs from before the fix date for anomalous process spawns, container escapes, or unexpected outbound network activity. 4) As a general control for any notebook-based SaaS platform, enforce strict sandbox/container isolation (gVisor, Firecracker, or equivalent), monitor kernel processes for host-breakout indicators, and limit blast radius by hard multi-tenant network segmentation.

What systems are affected by CVE-2021-42950?

This vulnerability affects the following AI/ML architecture patterns: training pipelines, notebook-based ML development environments, multi-tenant SaaS ML platforms.

What is the CVSS score for CVE-2021-42950?

No CVSS score has been assigned yet.

What is the AI security impact?

Affected AI Architectures

training pipelinesnotebook-based ML development environmentsmulti-tenant SaaS ML platforms

MITRE ATLAS Techniques

AML.T0050 Command and Scripting Interpreter

AML.T0105 Escape to Host

Compliance Controls Affected

EU AI Act: Article 15

NIST AI RMF: GOVERN 6.1

What are the technical details?

Original Advisory

Remote Code Execution (RCE) vulnerability exists in Zepl Notebooks all previous versions before October 25 2021. Users can register for an account and are allocated a set number of credits to try the product. Once users authenticate, they can proceed to create a new organization by which additional users can be added for various collaboration abilities, which allows malicious user to create new Zepl Notebooks with various languages, contexts, and deployment scenarios. Upon creating a new notebook with specially crafted malicious code, a user can then launch remote code execution.

Exploitation Scenario

An attacker signs up for a free Zepl trial account (no special privileges required), creates a new organization to unlock notebook creation, and then authors a notebook in a supported language (e.g., Python) containing code designed to escape the execution sandbox — for example, invoking OS-level system calls or exploiting a container/kernel weakness reachable from within the notebook runtime. On execution, the malicious code achieves remote code execution on the underlying host or container infrastructure, potentially allowing the attacker to enumerate or access other tenants' notebooks, stored datasets, or model artifacts hosted on the same shared platform.