CVE-2026-59215: Open WebUI IDOR leaks private

Q: Is CVE-2026-59215 actively exploited?

Proof-of-concept exploit code is publicly available for CVE-2026-59215, increasing the risk of exploitation.

Q: How to fix CVE-2026-59215?

Upgrade Open WebUI to version 0.10.0 or later, which fixes the missing channel-binding check on parent_id (see GHSA-73x5-h92w-xc2j and commit a66477b). Until patched, restrict shared/multi-tenant Open WebUI deployments to trusted user groups, avoid mixing high-sensitivity channels with broad-access accounts, and audit application logs for thread/reply requests referencing parent_id values outside a user's known channel membership. After patching, verify via changelog/version check that the fix is active, and consider periodic access-control regression testing on channel/thread endpoints given 130 other CVEs recorded in this package's history.

Q: What systems are affected by CVE-2026-59215?

This vulnerability affects the following AI/ML architecture patterns: self-hosted LLM chat UI, multi-tenant AI platforms, team collaboration/channel messaging within AI platforms.

Q: What is the CVSS score for CVE-2026-59215?

CVE-2026-59215 has a CVSS v3.1 base score of 3.1 (LOW). The EPSS exploitation probability is 0.32%.

CISO Take

Open WebUI, a widely self-hosted chat front-end for LLMs, shipped a broken access control in its channel/thread reply handling: the parent_id used to fetch thread context was never validated against the channel in the request URL, so any authenticated user could reference a message ID from a private or direct-message channel they don't belong to and pull back that thread's context. The blast radius is limited by design of the flaw itself — it requires an existing low-privileged account, high attack complexity (AC:H), and only confidentiality is affected (CVSS 3.1, no integrity or availability impact) — and there is no public exploit code, no Nuclei template, and it isn't in CISA KEV, with an EPSS score of 0.257% (still in the top 83rd percentile of scored CVEs, so not negligible). For teams running Open WebUI as a shared, multi-user deployment — common in enterprises centralizing LLM access — this is a realistic insider/lateral-disclosure risk: a curious or malicious internal user could read DM or private-channel content (which may include pasted credentials, internal prompts, or proprietary data) without needing external network access. Patch to Open WebUI 0.10.0 or later, which binds parent_id to its owning channel before returning thread data; until patched, review access logs for anomalous cross-channel thread fetches and restrict multi-user channel/DM usage to trusted user pools.

Sources: NVD GitHub Advisory EPSS

What is the risk?

Low overall severity (CVSS 3.1) driven by a narrow confidentiality-only impact and high attack complexity, but the vulnerability class — an IDOR/missing authorization check (CWE-639) — is a common and reliable exploitation pattern once understood. Requires an authenticated, low-privileged account (PR:L) and no user interaction, so any existing member of a shared instance is a potential attacker; there is no need for AI/ML domain expertise, just knowledge of the channel/thread API structure. No KEV listing, no public PoC, and a sub-1% EPSS score all point to low near-term exploitation likelihood, but the risk rises sharply for organizations running Open WebUI as shared multi-tenant infrastructure across teams with different sensitivity levels.

How does the attack unfold?

Valid Account Access

Attacker already holds a legitimate, low-privileged account on a shared Open WebUI instance.

AML.T0012

IDOR Request Crafting

Attacker submits a thread/reply request with a parent_id copied or guessed from another private or DM channel, exploiting the missing channel-binding check (CWE-639).

Cross-Channel Data Disclosure

The server returns thread context from the unauthorized channel, exposing potentially sensitive conversation content to the attacker.

AML.T0036

Valid Account Access

Attacker already holds a legitimate, low-privileged account on a shared Open WebUI instance.

AML.T0012

IDOR Request Crafting

Attacker submits a thread/reply request with a parent_id copied or guessed from another private or DM channel, exploiting the missing channel-binding check (CWE-639).

Cross-Channel Data Disclosure

The server returns thread context from the unauthorized channel, exposing potentially sensitive conversation content to the attacker.

AML.T0036

What systems are affected?

Package	Ecosystem	Vulnerable Range	Patched
Open WebUI	pip	< 0.10.0	`0.10.0`
149.6K 4 dependents Pushed 3d ago 81% patched ~6d to patch Full package profile →

Do you use Open WebUI? You're affected.

How severe is it?

CVSS 3.1

3.1 / 10

EPSS

0.3%

chance of exploitation in 30 days

Higher than 25% of all CVEs

Source: EPSS v3 — FIRST.org

Exploitation Status

Exploit Available

Exploitation: MEDIUM

Sophistication

Moderate

Exploitation Confidence

medium

○ CISA SSVC: Public PoC

○ Public PoC indexed (trickest/cve)

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?

AV Network

AC High

PR Low

UI None

S Unchanged

C Low

I None

A None

What should I do?

1 step

Upgrade Open WebUI to version 0.10.0 or later, which fixes the missing channel-binding check on parent_id (see GHSA-73x5-h92w-xc2j and commit a66477b). Until patched, restrict shared/multi-tenant Open WebUI deployments to trusted user groups, avoid mixing high-sensitivity channels with broad-access accounts, and audit application logs for thread/reply requests referencing parent_id values outside a user's known channel membership. After patching, verify via changelog/version check that the fix is active, and consider periodic access-control regression testing on channel/thread endpoints given 130 other CVEs recorded in this package's history.

What does CISA's SSVC say?

Decision Track*

Exploitation poc

Automatable No

Technical Impact partial

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

How is it classified?

Auth Bypass Data Leakage Framework AML.T0036 - Data from Information Repositories

Which compliance frameworks are affected?

This CVE is relevant to:

EU AI Act

Article 15 - Accuracy, robustness and cybersecurity

ISO 42001

A.6 (Data for AI systems) - Data management and access control

NIST AI RMF

MANAGE 4.1 - AI risks and benefits are monitored and managed

OWASP LLM Top 10

LLM02 - Sensitive Information Disclosure

Frequently Asked Questions

What is CVE-2026-59215?

Open WebUI, a widely self-hosted chat front-end for LLMs, shipped a broken access control in its channel/thread reply handling: the parent_id used to fetch thread context was never validated against the channel in the request URL, so any authenticated user could reference a message ID from a private or direct-message channel they don't belong to and pull back that thread's context. The blast radius is limited by design of the flaw itself — it requires an existing low-privileged account, high attack complexity (AC:H), and only confidentiality is affected (CVSS 3.1, no integrity or availability impact) — and there is no public exploit code, no Nuclei template, and it isn't in CISA KEV, with an EPSS score of 0.257% (still in the top 83rd percentile of scored CVEs, so not negligible). For teams running Open WebUI as a shared, multi-user deployment — common in enterprises centralizing LLM access — this is a realistic insider/lateral-disclosure risk: a curious or malicious internal user could read DM or private-channel content (which may include pasted credentials, internal prompts, or proprietary data) without needing external network access. Patch to Open WebUI 0.10.0 or later, which binds parent_id to its owning channel before returning thread data; until patched, review access logs for anomalous cross-channel thread fetches and restrict multi-user channel/DM usage to trusted user pools.

Is CVE-2026-59215 actively exploited?

Proof-of-concept exploit code is publicly available for CVE-2026-59215, increasing the risk of exploitation.

How to fix CVE-2026-59215?

Upgrade Open WebUI to version 0.10.0 or later, which fixes the missing channel-binding check on parent_id (see GHSA-73x5-h92w-xc2j and commit a66477b). Until patched, restrict shared/multi-tenant Open WebUI deployments to trusted user groups, avoid mixing high-sensitivity channels with broad-access accounts, and audit application logs for thread/reply requests referencing parent_id values outside a user's known channel membership. After patching, verify via changelog/version check that the fix is active, and consider periodic access-control regression testing on channel/thread endpoints given 130 other CVEs recorded in this package's history.

What systems are affected by CVE-2026-59215?

This vulnerability affects the following AI/ML architecture patterns: self-hosted LLM chat UI, multi-tenant AI platforms, team collaboration/channel messaging within AI platforms.

What is the CVSS score for CVE-2026-59215?

CVE-2026-59215 has a CVSS v3.1 base score of 3.1 (LOW). The EPSS exploitation probability is 0.32%.

What is the AI security impact?

Affected AI Architectures

self-hosted LLM chat UImulti-tenant AI platformsteam collaboration/channel messaging within AI platforms

MITRE ATLAS Techniques

AML.T0036 Data from Information Repositories

Compliance Controls Affected

EU AI Act: Article 15

ISO 42001: A.6 (Data for AI systems)

NIST AI RMF: MANAGE 4.1

OWASP LLM Top 10: LLM02

What are the technical details?

Original Advisory

Open WebUI is an extensible, feature-rich, and user-friendly self-hosted AI platform. Prior to 0.10.0, channel thread parent and reply handling did not bind parent_id to the channel in the URL, allowing an authenticated user to reference a message from another private or DM channel and disclose thread context across channels. This issue is fixed in version 0.10.0.

Exploitation Scenario

An organization runs a shared Open WebUI instance so multiple teams can access various LLMs through one interface, using channels and DMs to segment sensitive discussions (e.g., a security team's private channel vs. a general engineering channel). A low-privileged employee with only general-channel access observes or guesses a message/parent_id from the security team's private channel — for example via UI enumeration, shared links, or sequential ID patterns — and submits a thread/reply request referencing that parent_id from within a channel URL they do have access to. Because the server never validated that the parent_id actually belongs to the requesting channel, it returns the private thread's context, disclosing potentially sensitive content (credentials pasted for debugging, internal risk discussions, proprietary prompts) to an unauthorized internal user.

Weaknesses (CWE)

CWE-639 Authorization Bypass Through User-Controlled Key Primary CWE-639 Authorization Bypass Through User-Controlled Key

CWE-639 — Authorization Bypass Through User-Controlled Key: The system's authorization functionality does not prevent one user from gaining access to another user's data or record by modifying the key value identifying the data.

[Architecture and Design] For each and every data access, ensure that the user has sufficient privilege to access the record that is being requested.
[Architecture and Design, Implementation] Make sure that the key that is used in the lookup of a specific user's record is not controllable externally by the user or that any tampering can be detected.

Source: MITRE CWE corpus.