CVE-2026-25592: Semantic Kernel Path Traversal

Q: Is CVE-2026-25592 actively exploited?

No confirmed active exploitation of CVE-2026-25592 has been reported, but organizations should still patch proactively.

Q: How to fix CVE-2026-25592?

1) PATCH: Upgrade Microsoft.SemanticKernel.Core to 1.71.0+ (NuGet) and semantic-kernel to 1.39.3+ (pip) immediately — no exceptions for production systems. 2) WORKAROUND: If patching is delayed, implement a Function Invocation Filter that validates localFilePath against an explicit allowlist before any DownloadFileAsync or UploadFileAsync invocation — this is the vendor-documented compensating control. 3) DETECT: Alert on file writes outside designated agent working directories; monitor for unexpected modifications to /etc/, ~/.ssh/, cron directories, or application config paths from the Semantic Kernel process user. 4) AUDIT: Enumerate all usages of SessionsPythonPlugin in your codebase and CI/CD pipelines; verify dependency versions across all environments. 5) HARDEN: Run Semantic Kernel processes under a least-privilege OS user and within a restricted filesystem namespace — read-only container mounts where possible, seccomp profiles restricting filesystem syscalls.

Q: What systems are affected by CVE-2026-25592?

This vulnerability affects the following AI/ML architecture patterns: agent frameworks, plugin systems, code interpretation sandboxes, AI copilot backends, multi-agent orchestration systems.

Q: What is the CVSS score for CVE-2026-25592?

CVE-2026-25592 has a CVSS v3.1 base score of 9.9 (CRITICAL). The EPSS exploitation probability is 1.95%.

CISO Take

Patch now if you're running Semantic Kernel agents with SessionsPythonPlugin — this is a CVSS 9.9 arbitrary file write reachable by any authenticated user over the network with zero user interaction. The scope change (S:C) in the CVSS vector means successful exploitation can break container isolation, turning a compromised plugin into full host access. If immediate patching to 1.71.0 (.NET) / 1.39.3 (Python) is not possible, implement a Function Invocation Filter to allowlist valid file paths as the vendor-recommended workaround.

What is the risk?

Critical risk for organizations deploying Semantic Kernel-based AI agents. Network-accessible, low privilege required, no user interaction, scope change — this is close to a worst-case CVSS profile. EPSS (0.058%) indicates no active exploitation observed yet, but path traversal primitives in file I/O plugins are trivial to weaponize once the target API is identified. Risk is amplified in multi-tenant SaaS products built on Semantic Kernel where untrusted users have agent access.

What systems are affected?

Package	Ecosystem	Vulnerable Range	Patched
Semantic Kernel	pip	< 1.39.3	`1.39.3`
28.2K 27 dependents Pushed 3d ago 100% patched ~7d to patch Full package profile →
Microsoft.SemanticKernel.Core	nuget	< 1.71.0	`1.71.0`

How severe is it?

CVSS 3.1

9.9 / 10

EPSS

1.9%

chance of exploitation in 30 days

Higher than 78% of all CVEs

Source: EPSS v3 — FIRST.org

Exploitation Status

No known exploitation

Sophistication

Trivial

What is the attack surface?

AV Network

AC Low

PR Low

UI None

S Changed

C High

I High

A High

What should I do?

1 step

1) PATCH: Upgrade Microsoft.SemanticKernel.Core to 1.71.0+ (NuGet) and semantic-kernel to 1.39.3+ (pip) immediately — no exceptions for production systems. 2) WORKAROUND: If patching is delayed, implement a Function Invocation Filter that validates localFilePath against an explicit allowlist before any DownloadFileAsync or UploadFileAsync invocation — this is the vendor-documented compensating control. 3) DETECT: Alert on file writes outside designated agent working directories; monitor for unexpected modifications to /etc/, ~/.ssh/, cron directories, or application config paths from the Semantic Kernel process user. 4) AUDIT: Enumerate all usages of SessionsPythonPlugin in your codebase and CI/CD pipelines; verify dependency versions across all environments. 5) HARDEN: Run Semantic Kernel processes under a least-privilege OS user and within a restricted filesystem namespace — read-only container mounts where possible, seccomp profiles restricting filesystem syscalls.

What does CISA's SSVC say?

Decision Track

Exploitation none

Automatable No

Technical Impact total

Source: CISA Vulnrichment (SSVC v2.0). Decision based on the CISA Coordinator decision tree.

How is it classified?

Code Execution Supply Chain Data Extraction Framework Agent Plugin AML.T0049 - Exploit Public-Facing Application AML.T0051.001 - Indirect AML.T0053 - AI Agent Tool Invocation AML.T0086 - Exfiltration via AI Agent Tool Invocation AML.T0101 - Data Destruction via AI Agent Tool Invocation AML.T0105 - Escape to Host

Which compliance frameworks are affected?

This CVE is relevant to:

EU AI Act

Art. 15 - Accuracy, Robustness and Cybersecurity

ISO 42001

8.5 - AI System Security A.6.1.2 - AI System Security Controls

NIST AI RMF

GOVERN-1.7 - AI Risk Management — Third-Party Components MANAGE-2.2 - Risk Treatment for AI Systems

OWASP LLM Top 10

LLM06 - Excessive Agency LLM07 - Insecure Plugin Design LLM08 - Excessive Agency

Frequently Asked Questions

What is CVE-2026-25592?

Patch now if you're running Semantic Kernel agents with SessionsPythonPlugin — this is a CVSS 9.9 arbitrary file write reachable by any authenticated user over the network with zero user interaction. The scope change (S:C) in the CVSS vector means successful exploitation can break container isolation, turning a compromised plugin into full host access. If immediate patching to 1.71.0 (.NET) / 1.39.3 (Python) is not possible, implement a Function Invocation Filter to allowlist valid file paths as the vendor-recommended workaround.

Is CVE-2026-25592 actively exploited?

No confirmed active exploitation of CVE-2026-25592 has been reported, but organizations should still patch proactively.

How to fix CVE-2026-25592?

1) PATCH: Upgrade Microsoft.SemanticKernel.Core to 1.71.0+ (NuGet) and semantic-kernel to 1.39.3+ (pip) immediately — no exceptions for production systems. 2) WORKAROUND: If patching is delayed, implement a Function Invocation Filter that validates localFilePath against an explicit allowlist before any DownloadFileAsync or UploadFileAsync invocation — this is the vendor-documented compensating control. 3) DETECT: Alert on file writes outside designated agent working directories; monitor for unexpected modifications to /etc/, ~/.ssh/, cron directories, or application config paths from the Semantic Kernel process user. 4) AUDIT: Enumerate all usages of SessionsPythonPlugin in your codebase and CI/CD pipelines; verify dependency versions across all environments. 5) HARDEN: Run Semantic Kernel processes under a least-privilege OS user and within a restricted filesystem namespace — read-only container mounts where possible, seccomp profiles restricting filesystem syscalls.

What systems are affected by CVE-2026-25592?

This vulnerability affects the following AI/ML architecture patterns: agent frameworks, plugin systems, code interpretation sandboxes, AI copilot backends, multi-agent orchestration systems.

What is the CVSS score for CVE-2026-25592?

CVE-2026-25592 has a CVSS v3.1 base score of 9.9 (CRITICAL). The EPSS exploitation probability is 1.95%.

What is the AI security impact?

Affected AI Architectures

agent frameworksplugin systemscode interpretation sandboxesAI copilot backendsmulti-agent orchestration systems

MITRE ATLAS Techniques

AML.T0049 Exploit Public-Facing Application

AML.T0051.001 Indirect

AML.T0053 AI Agent Tool Invocation

AML.T0086 Exfiltration via AI Agent Tool Invocation

AML.T0101 Data Destruction via AI Agent Tool Invocation

AML.T0105 Escape to Host

Compliance Controls Affected

EU AI Act: Art. 15

ISO 42001: 8.5, A.6.1.2

NIST AI RMF: GOVERN-1.7, MANAGE-2.2

OWASP LLM Top 10: LLM06, LLM07, LLM08

What are the technical details?

Original Advisory

Semantic Kernel is an SDK used to build, orchestrate, and deploy AI agents and multi-agent systems. Prior to 1.71.0, an Arbitrary File Write vulnerability has been identified in Microsoft's Semantic Kernel .NET SDK, specifically within the SessionsPythonPlugin. The problem has been fixed in Microsoft.SemanticKernel.Core version 1.71.0. As a mitigation, users can create a Function Invocation Filter which checks the arguments being passed to any calls to DownloadFileAsync  or UploadFileAsync and ensures the provided localFilePath is allow listed.

Exploitation Scenario

An adversary with low-privilege access to a Semantic Kernel-backed AI coding assistant — a trial user, a compromised internal account, or an attacker who has obtained a valid API token — sends a crafted agent invocation calling DownloadFileAsync with localFilePath set to '../../.ssh/authorized_keys' and attacker-controlled content as the file payload. The unpatched SessionsPythonPlugin writes the attacker's SSH public key to the host's authorized_keys file, granting persistent SSH access to the underlying infrastructure. In a cloud-hosted multi-tenant scenario, this single request escalates to full control of the agent execution environment and potential lateral movement to other tenant workloads. No UI interaction from a victim is required at any stage.

Weaknesses (CWE)

CWE-22 Improper Limitation of a Pathname to a Restricted Directory ('Path Traversal') Primary CWE-22 Improper Limitation of a Pathname to a Restricted Directory ('Path Traversal')

CWE-22 — Improper Limitation of a Pathname to a Restricted Directory ('Path Traversal'): The product uses external input to construct a pathname that is intended to identify a file or directory that is located underneath a restricted parent directory, but the product does not properly neutralize special elements within the pathname that can cause the pathname to resolve to a location that is outside of the restricted directory.

[Implementation] Assume all input is malicious. Use an "accept known good" input validation strategy, i.e., use a list of acceptable inputs that strictly conform to specifications. Reject any input that does not strictly conform to specifications, or transform it into something that does. When performing input validation, consider all potentially relevant properties, including length, type of input, the full range of acceptable values, missing or extra inputs, syntax, consistency across related fields, and conformance to business rules. As an example of business rule logic, "boat" may be syntactically valid because it only contains alphanumeric characters, but it is not valid if the input is only expected to contain colors such as "red" or "blue." Do not rely exclusively on looking for malicious or malformed inputs. This is likely to miss at least one undesirable input, especially if the code's environment changes. This can give attackers enough room to bypass the intended validation. However, denylis
[Architecture and Design] For any security checks that are performed on the client side, ensure that these checks are duplicated on the server side, in order to avoid CWE-602. Attackers can bypass the client-side checks by modifying values after the checks have been performed, or by changing the client to remove the client-side checks entirely. Then, these modified values would be submitted to the server.

Source: MITRE CWE corpus.