CVE-2026-59816: Joplin Server: path traversal in transcribe API leaks data

MEDIUMCISA: TRACK*

Published September 21, 2026

CISO Take

Joplin Server's transcription API (used when TRANSCRIBE_ENABLED=true) fails to sanitize the job ID, so an authenticated user can inject URL-encoded slashes and parent-directory segments to escape the intended /transcribe/ path and make the server proxy requests to other transcription-backend endpoints, returning their responses. The impact is limited to low-confidentiality disclosure of internal administrative, health, or configuration data (CVSS 4.3, no integrity or availability impact). Exploitation likelihood is low-to-moderate: EPSS is 0.36% (top 73%), there is no public exploit, no Nuclei template, and it is not in CISA KEV, with SSVC at Track*. Organizations running self-hosted Joplin Server with the transcription feature enabled should upgrade to 3.7.7 or later, and where patching is delayed, disable TRANSCRIBE_ENABLED and restrict access to the transcription backend to the Joplin Server only.

Sources:NVDEPSSGitHub AdvisoryATLAS

What is the risk?

Medium-to-low risk. The attack is network-reachable with low complexity and needs no user interaction, but it requires a valid authenticated account and the non-default TRANSCRIBE_ENABLED=true setting. Impact is confined to limited confidentiality loss from the proxied backend (server-side request path manipulation, SSRF-like behavior). There is no known exploitation, no public PoC, and no KEV listing. Risk rises where the transcription backend holds sensitive configuration or credentials, or where open registration lets untrusted users obtain accounts.

How does the attack unfold?

Initial Access

Attacker authenticates to a Joplin Server with TRANSCRIBE_ENABLED=true using a low-privilege account.

AML.T0012

Exploitation

Attacker sends a transcribe request whose job ID contains URL-encoded slashes and parent-directory segments, escaping the /transcribe/ prefix.

AML.T0049

Collection

The server proxies the request to other transcription-backend endpoints and returns their responses, exposing admin, health, or configuration data.

AML.T0037

Impact

Leaked internal details help the attacker map the backend AI service and plan follow-on attacks.

Initial Access

Attacker authenticates to a Joplin Server with TRANSCRIBE_ENABLED=true using a low-privilege account.

AML.T0012

Exploitation

Attacker sends a transcribe request whose job ID contains URL-encoded slashes and parent-directory segments, escaping the /transcribe/ prefix.

AML.T0049

Collection

The server proxies the request to other transcription-backend endpoints and returns their responses, exposing admin, health, or configuration data.

AML.T0037

Impact

Leaked internal details help the attacker map the backend AI service and plan follow-on attacks.

How severe is it?

CVSS 3.1

4.3 / 10

EPSS

0.4%

chance of exploitation in 30 days

Higher than 29% of all CVEs

Source: EPSS v3 — FIRST.org

Exploitation Status

Exploit Available

Exploitation: MEDIUM

Sophistication

Trivial

Exploitation Confidence

medium

○CISA SSVC: Public PoC

Composite signal derived from CISA KEV, VulnCheck KEV, CISA SSVC, EPSS, Metasploit, Exploit-DB, trickest/cve, Nuclei templates, and inthewild.io exploitation reports.

What is the attack surface?

AVNetwork

ACLow

PRLow

UINone

SUnchanged

CLow

INone

ANone

What should I do?

1 step

1) Upgrade Joplin Server to 3.7.7 or later (fix commit d97a5091). 2) If you cannot patch immediately, set TRANSCRIBE_ENABLED=false. 3) Isolate the transcription backend on a private network, allow only the Joplin Server to reach it, and require authentication on its admin and health endpoints. 4) Review server access logs for requests to /api/transcribe/ containing %2f, %2e%2e, or ../ sequences. 5) Restrict or review user registration on Joplin Server, and rotate any secrets exposed through the backend configuration endpoints if suspicious requests are found.

What does CISA's SSVC say?

DecisionTrack*

Exploitationpoc

AutomatableNo

Technical Impactpartial

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

How is it classified?

Data LeakageData ExtractionAPIInferenceAML.T0012- Valid Accounts AML.T0037- Data from Local System AML.T0049- Exploit Public-Facing Application

Which compliance frameworks are affected?

This CVE is relevant to:

EU AI Act

Art. 15-Accuracy, robustness and cybersecurity

ISO 42001

A.6.2.6-AI system operation and monitoring

NIST AI RMF

MANAGE 2.3-Procedures to respond to and recover from previously unknown risks

OWASP LLM Top 10

LLM06-Sensitive Information Disclosure

Frequently Asked Questions

What is CVE-2026-59816?

Is CVE-2026-59816 actively exploited?

No confirmed active exploitation of CVE-2026-59816 has been reported, but organizations should still patch proactively.

How to fix CVE-2026-59816?

1) Upgrade Joplin Server to 3.7.7 or later (fix commit d97a5091). 2) If you cannot patch immediately, set TRANSCRIBE_ENABLED=false. 3) Isolate the transcription backend on a private network, allow only the Joplin Server to reach it, and require authentication on its admin and health endpoints. 4) Review server access logs for requests to /api/transcribe/ containing %2f, %2e%2e, or ../ sequences. 5) Restrict or review user registration on Joplin Server, and rotate any secrets exposed through the backend configuration endpoints if suspicious requests are found.

What systems are affected by CVE-2026-59816?

This vulnerability affects the following AI/ML architecture patterns: Self-hosted AI transcription services, API gateways proxying to ML backends, Note-taking platforms with AI features.

What is the CVSS score for CVE-2026-59816?

CVE-2026-59816 has a CVSS v3.1 base score of 4.3 (MEDIUM). The EPSS exploitation probability is 0.36%.

What is the AI security impact?

Affected AI Architectures

Self-hosted AI transcription servicesAPI gateways proxying to ML backendsNote-taking platforms with AI features

MITRE ATLAS Techniques

AML.T0012Valid Accounts

AML.T0037Data from Local System

AML.T0049Exploit Public-Facing Application

Compliance Controls Affected

EU AI Act: Art. 15

ISO 42001: A.6.2.6

NIST AI RMF: MANAGE 2.3

OWASP LLM Top 10: LLM06

What are the technical details?

Original Advisory

Joplin is an open source note-taking and to-do application that organises notes and lists into notebooks. Prior to 3.7.7, the GET /api/transcribe/:id and POST /api/transcribe/:id handlers in packages/server/src/routes/api/transcribe.ts on Joplin Server instances with TRANSCRIBE_ENABLED=true pass the decoded id directly into transcription backend URLs. An authenticated user can place URL-encoded slash and parent-directory segments in the job ID, causing parseSubPath() to decode a path that escapes the intended /transcribe/ prefix. The server then proxies the request to other transcription-backend endpoints and can return their responses, potentially exposing internal administrative, health, or configuration data. This issue is fixed in version 3.7.7.

Exploitation Scenario

An attacker with a low-privilege account on a Joplin Server (for example through open registration or a compromised user credential) sends GET /api/transcribe/..%2F..%2Fadmin%2Fconfig. The server decodes the ID in parseSubPath(), the path escapes the /transcribe/ prefix, and the request is proxied to an internal transcription-backend endpoint. The response, which may contain health status, version details, or configuration values such as model settings or internal addresses, is returned to the attacker. They use this to map the internal AI service and prepare further attacks.

Weaknesses (CWE)

CWE-22Improper Limitation of a Pathname to a Restricted DirectoryPrimary CWE-22Improper Limitation of a Pathname to a Restricted Directory

CWE-22 — Improper Limitation of a Pathname to a Restricted Directory: 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.