CVE-2026-17631: Langflow authenticated SSRF exposes

CISO Take

IBM Langflow OSS versions 1.0.0 through 1.10.2 contain a server-side request forgery flaw (CWE-918) that lets any authenticated user, even one with low privileges, coerce the backend into making requests to internal or otherwise restricted network destinations. This matters because Langflow is a visual LLM/agent orchestration tool typically deployed with network access to internal services, secrets managers, and cloud metadata endpoints — exactly the kind of blast radius SSRF is built to abuse. There is no evidence of active exploitation: it's not in CISA KEV, no EPSS score or public exploit code exists, and no Nuclei template has been published, so treat this as a proactive patch rather than an emergency. The CVSS 5.0 (network, low complexity, low privileges, confidentiality-low, scope-changed) reflects modest direct impact but the scope-changed flag signals it can reach beyond the vulnerable component itself. Patch to the fixed IBM Langflow release once available, restrict outbound network egress from Langflow hosts (especially to 169.254.169.254 and internal admin APIs), and audit any flow components that accept user-supplied URLs.

Sources:NVDibm.comATLAS

What is the risk?

Medium severity per CVSS (5.0) but the practical risk depends heavily on deployment context. The requirement for authentication (PR:L) meaningfully reduces the exploitable population versus an unauthenticated SSRF, and the confidentiality impact is rated low with no integrity or availability impact. However, SSRF vulnerabilities in AI orchestration platforms are historically underrated because the real damage comes from what the SSRF can reach — cloud metadata services, internal APIs, or credentials embedded in the deployment — rather than the CVSS score itself. No EPSS data, no KEV listing, and no public exploit or scanner template currently exist, so near-term mass exploitation is unlikely. Organizations running multi-tenant or externally-accessible Langflow instances with loosely-scoped user roles carry elevated risk.

How does the attack unfold?

Authenticated Access

Attacker obtains or already holds a low-privileged, valid Langflow user account.

AML.T0012

Malicious Flow Construction

Attacker creates or edits a flow component (e.g., HTTP request or webhook node) and sets its target URL to an internal or restricted address.

AML.T0049

Server-Side Request Execution

Langflow's backend issues the outbound request on the attacker's behalf without validating the destination, exploiting CWE-918.

Sensitive Data Disclosure

The response from the internal target (e.g., cloud metadata, internal API) is returned in the flow output, leaking sensitive information to the attacker.

Authenticated Access

Attacker obtains or already holds a low-privileged, valid Langflow user account.

AML.T0012

Malicious Flow Construction

Attacker creates or edits a flow component (e.g., HTTP request or webhook node) and sets its target URL to an internal or restricted address.

AML.T0049

Server-Side Request Execution

Langflow's backend issues the outbound request on the attacker's behalf without validating the destination, exploiting CWE-918.

Sensitive Data Disclosure

The response from the internal target (e.g., cloud metadata, internal API) is returned in the flow output, leaking sensitive information to the attacker.

What systems are affected?

Package	Ecosystem	Vulnerable Range	Patched
Langflow	pip	—	No patch
156.2KPushed today18% patched~66d to patchFull package profile →

Do you use Langflow? You're affected.

How severe is it?

CVSS 3.1

6.5 / 10

EPSS

0.4%

chance of exploitation in 30 days

Higher than 35% of all CVEs

Source: EPSS v3 — FIRST.org

Exploitation Status

No known exploitation

Sophistication

Moderate

What is the attack surface?

AVNetwork

ACLow

PRLow

UINone

SUnchanged

CHigh

INone

ANone

What should I do?

1 step

Upgrade to the IBM Langflow patched release above 1.10.2 as soon as it is published (monitor the IBM Support advisory at the referenced URL). Until patched, restrict egress network access from the Langflow host/container to only what workflows legitimately require, and explicitly block access to cloud metadata IPs (169.254.169.254, fd00:ec2::254) via network policy, iptables, or IMDSv2 enforcement. Review role-based access controls so low-privilege users cannot create or edit flow components that make arbitrary outbound HTTP requests. Add egress allowlisting/validation on any custom HTTP-request or webhook node types. For detection, monitor outbound connections from Langflow hosts for requests to internal RFC1918 ranges, link-local metadata addresses, or unexpected internal hostnames, and review Langflow flow definitions for newly added or modified HTTP/URL-based components.

What does CISA's SSVC say?

DecisionTrack

Exploitationnone

AutomatableNo

Technical Impactpartial

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

How is it classified?

Data LeakagePrivacy ViolationFrameworkAgentAML.T0049- Exploit Public-Facing Application AML.T0075- Cloud Service Discovery

Which compliance frameworks are affected?

This CVE is relevant to:

EU AI Act

Article 15-Accuracy, robustness and cybersecurity

OWASP LLM Top 10

LLM07:2025-Insecure Plugin Design

Frequently Asked Questions

What is CVE-2026-17631?

IBM Langflow OSS versions 1.0.0 through 1.10.2 contain a server-side request forgery flaw (CWE-918) that lets any authenticated user, even one with low privileges, coerce the backend into making requests to internal or otherwise restricted network destinations. This matters because Langflow is a visual LLM/agent orchestration tool typically deployed with network access to internal services, secrets managers, and cloud metadata endpoints — exactly the kind of blast radius SSRF is built to abuse. There is no evidence of active exploitation: it's not in CISA KEV, no EPSS score or public exploit code exists, and no Nuclei template has been published, so treat this as a proactive patch rather than an emergency. The CVSS 5.0 (network, low complexity, low privileges, confidentiality-low, scope-changed) reflects modest direct impact but the scope-changed flag signals it can reach beyond the vulnerable component itself. Patch to the fixed IBM Langflow release once available, restrict outbound network egress from Langflow hosts (especially to 169.254.169.254 and internal admin APIs), and audit any flow components that accept user-supplied URLs.

Is CVE-2026-17631 actively exploited?

No confirmed active exploitation of CVE-2026-17631 has been reported, but organizations should still patch proactively.

How to fix CVE-2026-17631?

Upgrade to the IBM Langflow patched release above 1.10.2 as soon as it is published (monitor the IBM Support advisory at the referenced URL). Until patched, restrict egress network access from the Langflow host/container to only what workflows legitimately require, and explicitly block access to cloud metadata IPs (169.254.169.254, fd00:ec2::254) via network policy, iptables, or IMDSv2 enforcement. Review role-based access controls so low-privilege users cannot create or edit flow components that make arbitrary outbound HTTP requests. Add egress allowlisting/validation on any custom HTTP-request or webhook node types. For detection, monitor outbound connections from Langflow hosts for requests to internal RFC1918 ranges, link-local metadata addresses, or unexpected internal hostnames, and review Langflow flow definitions for newly added or modified HTTP/URL-based components.

What systems are affected by CVE-2026-17631?

This vulnerability affects the following AI/ML architecture patterns: agent frameworks, LLM orchestration pipelines, tool/plugin integrations.

What is the CVSS score for CVE-2026-17631?

CVE-2026-17631 has a CVSS v3.1 base score of 6.5 (MEDIUM). The EPSS exploitation probability is 0.40%.

What is the AI security impact?

Affected AI Architectures

agent frameworksLLM orchestration pipelinestool/plugin integrations

MITRE ATLAS Techniques

AML.T0049Exploit Public-Facing Application

AML.T0075Cloud Service Discovery

Compliance Controls Affected

EU AI Act: Article 15

OWASP LLM Top 10: LLM07:2025

What are the technical details?

Original Advisory

IBM Langflow OSS 1.0.0 through 1.10.2 could allow a remote authenticated attacker to obtain sensitive information due to a server-side request forgery (SSRF) vulnerability.

Exploitation Scenario

An attacker obtains or is granted a low-privileged Langflow account (e.g., a limited flow-builder role in a multi-tenant deployment). They create or edit a flow using a component that accepts a URL parameter — such as an HTTP request node or webhook integration — and point it at an internal-only address, for example the cloud provider's instance metadata endpoint or an internal admin API with no external exposure. Because Langflow's backend does not adequately restrict destination targets for these requests, it issues the request on the attacker's behalf and returns the response within the flow output, potentially leaking temporary cloud credentials, internal service banners, or configuration data the attacker can use to pivot further into the environment.

Weaknesses (CWE)

CWE-918Server-Side Request ForgeryPrimary CWE-918Server-Side Request Forgery

CWE-918 — Server-Side Request Forgery: The web server receives a URL or similar request from an upstream component and retrieves the contents of this URL, but it does not sufficiently ensure that the request is being sent to the expected destination.

[Implementation, Operation] If there is a known list of applications, IP, hostname, and/or domains to which the product is allowed to communicate, then use allowlists to ensure that outgoing requests are restricted to these known entities. The allowlists could be in the application code or possibly enforced by firewalls or network segmentation [REF-1511].
[Implementation, Operation] If the product is expected to be able to connect to arbitrary IP addresses or domains, consider using a denylist that includes private network addresses, internal domain names, etc. The denylist could be in the application code or possibly enforced by firewalls or network segmentation [REF-1511].

Source: MITRE CWE corpus.