In version v0.3.8 of open-webui, an improper privilege management vulnerability exists in the API endpoints GET /api/v1/documents/ and POST /rag/api/v1/doc. This vulnerability allows a lower-privileged user to access and overwrite files managed by a higher-privileged admin. By exploiting this vulnerability, an attacker can view metadata of files uploaded by an admin and overwrite these files, compromising the integrity and availability of the RAG models.

Open WebUI is a self-hosted artificial intelligence platform designed to operate entirely offline. Prior to 0.9.0, the channel router does not call filter_allowed_access_grants on either create or update paths. A non-admin user who can create group channels (or who owns a channel) can submit arbitrary access grants — including public wildcard grants — and those grants are stored verbatim, bypassing the admin's permission framework. This vulnerability is fixed in 0.9.0.

Open WebUI is a self-hosted artificial intelligence platform designed to operate entirely offline. Prior to 0.8.11, an internal-only bypass_filter parameter is exposed on the /openai/chat/completions and /ollama/api/chat HTTP endpoints via FastAPI query string binding, allowing any authenticated user to append ?bypass_filter=true and bypass model access control checks to invoke admin-restricted models. This vulnerability is fixed in 0.8.11.

In version v0.3.8 of open-webui/open-webui, a vulnerability exists where a token is returned when a user with a pending role logs in. This allows the user to perform actions without admin confirmation, bypassing the intended approval process.

Open WebUI is a self-hosted artificial intelligence platform designed to operate entirely offline. Prior to 0.9.3, the POST /api/v1/notes/{id}/pin endpoint performs a write operation (toggling the is_pinned field) but only checks for read permission. Users with read-only access to a shared note can pin/unpin it, which is a state-modifying action that should require write permission. This vulnerability is fixed in 0.9.3.

Open WebUI is an extensible, feature-rich, and user-friendly self-hosted AI platform. From 0.9.0 before 0.10.0, execute_automation rehydrated automation owners without rechecking that they were still active or still had features.automations, and check_model_access only enforced private-model grants for the exact user role, allowing deactivated pending users to continue scheduled model execution. This issue is fixed in version 0.10.0.

Open WebUI is an extensible, feature-rich, and user-friendly self-hosted AI platform. From 0.9.6 before 0.10.0, _verify_knowledge_file_access only checked read access while file write and delete routes later trusted object-derived access through writable model meta.knowledge entries, allowing a user with read-only knowledge file access to upgrade to file write or delete operations. This issue is fixed in version 0.10.0.

Open WebUI is an extensible, feature-rich, and user-friendly self-hosted AI platform. From 0.8.11 before 0.10.0, POST /api/v1/images/edit required only a verified account and did not enforce the global image-edit switch or the per-user image-generation permission, allowing a non-admin user to invoke server-side image editing with administrator-configured provider credentials. This issue is fixed in version 0.10.0.

Open WebUI is a self-hosted artificial intelligence platform designed to operate entirely offline. Prior to 0.8.11, the ydoc:document:join Socket.IO handler checks note ownership only when the document_id starts with note: (colon). However, the YdocManager storage layer normalizes all document IDs by replacing colons with underscores (document_id.replace(":", "_")). An attacker can join a document room using note_<id> (underscore) instead of note:<id> (colon), bypassing the authorization check entirely while accessing the same underlying Yjs document. The server then returns the full document state, leaking the victim's private note contents. This vulnerability is fixed in 0.8.11.

Open WebUI is a self-hosted artificial intelligence platform designed to operate entirely offline. Prior to 0.9.6, several direct, index-addressed Ollama proxy routes accept a caller-supplied url_idx path parameter and use it as a raw index into the admin-configured OLLAMA_BASE_URLS list. Access control on these routes validates only whether the user may use the requested model, never which backend the request is routed to. Any authenticated user can append an arbitrary url_idx to force their request onto an Ollama backend they were never authorized to reach, including internal, higher-privilege, or explicitly admin-disabled backends. This vulnerability is fixed in 0.9.6.

Open WebUI is a self-hosted artificial intelligence platform designed to operate entirely offline. Prior to 0.8.12, the /api/v1/utils/code/execute endpoint executes arbitrary Python code via Jupyter for any verified user, even when the admin has set ENABLE_CODE_EXECUTION=false. The feature gate is not enforced on the API endpoint — the configuration says "disabled" but code still executes. This vulnerability is fixed in 0.8.12.

Open WebUI is a self-hosted artificial intelligence platform designed to operate entirely offline. Prior to 0.9.0, Open WebUI allows admins to restrict which API endpoints an API key can access. When an API key is restricted from /api/v1/messages, requests using the Authorization: Bearer sk-... header are correctly blocked with 403. However, the same key sent via the x-api-key header bypasses the restriction entirely — the request is authenticated, the model is invoked, and a full response is returned. This vulnerability is fixed in 0.9.0.

Open WebUI is a self-hosted artificial intelligence platform designed to operate entirely offline. Prior to 0.9.0, the _validate_collection_access function uses an incomplete allowlist that only enforces ownership checks for collections matching user-memory-* and file-* patterns. All other collection names pass through unchecked — including the system-level knowledge-bases meta-collection, which stores the IDs, names, and descriptions of every knowledge base on the instance. Any authenticated user can query this meta-collection directly via the retrieval query endpoints to obtain a global index of all knowledge bases across all users. This vulnerability is fixed in 0.9.0.

Open WebUI is a self-hosted artificial intelligence platform designed to operate entirely offline. Prior to 0.9.0, the ydoc:document:update Socket.IO event handler checks whether the sender is a member of the document's Socket.IO room (line 678) but does not verify that the sender has write permission. Users with read-only access join the document room via ydoc:document:join, which only requires read permission (line 520). Once in the room, the user can emit ydoc:document:update events that modify the in-memory Yjs document state and are broadcast to all other collaborators in real time. This vulnerability is fixed in 0.9.0.

Open WebUI is an extensible, feature-rich, and user-friendly self-hosted AI platform. Prior to 0.10.0, the file upload path accepted metadata.knowledge_id and auto-linked uploaded files to a target knowledge base without applying the write-access check used by /api/v1/knowledge//file/add, allowing read-only knowledge-base users to add arbitrary files. This issue is fixed in version 0.10.0.

Open WebUI is a self-hosted artificial intelligence platform designed to operate entirely offline. Prior to 0.1.124, the API does not properly validate that the user has an authorized user role of user. By default, when Open WebUI is configured with new sign-ups enabled, the default user role is set to pending. In this configuration, an administrator is required to go into the Admin management panel following a new user registration and reconfigure the user to have a role of either user or admin before that user is able to access the web application. This vulnerability is fixed in 0.1.124.

Mattermost fails to disable public Boards after the "Enable Publicly-Shared Boards" configuration option is disabled, resulting in previously-shared public Boards to remain accessible.

Incomplete authorization of linked device synchronization messages in WhatsApp for iOS prior to v2.25.21.73, WhatsApp Business for iOS v2.25.21.78, and WhatsApp for Mac v2.25.21.78 could have allowed an unrelated user to trigger processing of content from an arbitrary URL on a target’s device. We assess that this vulnerability, in combination with an OS-level vulnerability on Apple platforms (CVE-2025-43300), may have been exploited in a sophisticated attack against specific targeted users.

An issue has been discovered in GitLab affecting all versions before 16.7.6, all versions starting from 16.8 before 16.8.3, all versions starting from 16.9 before 16.9.1. It was possible for group members with sub-maintainer role to change the title of privately accessible deploy keys associated with projects in the group.

Zoho ManageEngine ServiceDesk Plus before 14202, ServiceDesk Plus MSP before 14300, and SupportCenter Plus before 14300 have a privilege escalation vulnerability in the Release module that allows unprivileged users to access the Reminders of a release ticket and make modifications.

Synapse is a Matrix protocol homeserver written in Python with the Twisted framework. A discovered oEmbed or image URL can bypass the `url_preview_url_blacklist` setting potentially allowing server side request forgery or bypassing network policies. Impact is limited to IP addresses allowed by the `url_preview_ip_range_blacklist` setting (by default this only allows public IPs) and by the limited information returned to the client: 1. For discovered oEmbed URLs, any non-JSON response or a JSON response which includes non-oEmbed information is discarded. 2. For discovered image URLs, any non-image response is discarded. Systems which have URL preview disabled (via the `url_preview_enabled` setting) or have not configured a `url_preview_url_blacklist` are not affected. This issue has been addressed in version 1.85.0. Users are advised to upgrade. User unable to upgrade may also disable URL previews.

Directus is a real-time API and App dashboard for managing SQL database content. In affected versions if there are two overlapping policies for the `update` action that allow access to different fields, instead of correctly checking access permissions against the item they apply for the user is allowed to update the superset of fields allowed by any of the policies. E.g. have one policy allowing update access to `field_a` if the `id == 1` and one policy allowing update access to `field_b` if the `id == 2`. The user with both these policies is allowed to update both `field_a` and `field_b` for the items with ids `1` and `2`. Before v11, if a user was allowed to update an item they were allowed to update the fields that the single permission, that applied to that item, listed. With overlapping permissions this isn't as clear cut anymore and the union of fields might not be the fields the user is allowed to update for that specific item. The solution that this PR introduces is to evaluate the permissions for each field that the user tries to update in the validateItemAccess DB query, instead of only verifying access to the item as a whole. This is done by, instead of returning the actual field value, returning a flag that indicates if the user has access to that field. This uses the same case/when mechanism that is used for stripping out non permitted field that is at the core of the permissions engine. As a result, for every item that the access is validated for, the expected result is an item that has either 1 or null for all the "requested" fields instead of any of the actual field values. These results are not useful for anything other than verifying the field level access permissions. The final check in validateItemAccess can either fail if the number of items does not match the number of items the access is checked for (ie. the user does not have access to the item at all) or if not all of the passed in fields have access permissions for any of the returned items. This is a vulnerability that allows update access to unintended fields, potentially impacting the password field for user accounts. This has been addressed in version 11.1.2 and all users are advised to upgrade. There are no known workarounds for this vulnerability.

A security defect was identified in Foundry workspace-server that enabled a user to bypass an authorization check and view settings related to 'Developer Mode'. This enabled users with insufficient privilege the ability to view and interact with Developer Mode settings in a limited capacity. A fix was deployed with workspace-server 7.7.0.

Incorrect Authorization vulnerability in Apache Cassandra allowing users to access a datacenter or IP/CIDR groups they should not be able to when using CassandraNetworkAuthorizer or CassandraCIDRAuthorizer. Users with restricted data center access can update their own permissions via data control language (DCL) statements on affected versions. This issue affects Apache Cassandra: from 4.0.0 through 4.0.15 and from 4.1.0 through 4.1.7 for CassandraNetworkAuthorizer, and from 5.0.0 through 5.0.2 for both CassandraNetworkAuthorizer and CassandraCIDRAuthorizer. Operators using CassandraNetworkAuthorizer or CassandraCIDRAuthorizer on affected versions should review data access rules for potential breaches. Users are recommended to upgrade to versions 4.0.16, 4.1.8, 5.0.3, which fixes the issue.

XWiki Platform is a generic wiki platform. Starting in version 3.0-milestone-1, it's possible to execute a script with the right of another user, provided the target user does not have programming right. The problem has been patched in XWiki 14.8-rc-1, 14.4.5, and 13.10.10. There are no known workarounds for this issue.

GitLab has remediated an issue in GitLab CE/EE affecting all versions from 15.10 before 18.10.8, 18.11 before 18.11.5, and 19.0 before 19.0.2 that under certain conditions could have allowed an authenticated user with developer-role permissions to modify hidden merge requests due to incorrect authorization enforcements.

A flaw was found in KubeVirt's Role-Based Access Control (RBAC) evaluation logic. The authorization mechanism improperly truncates subresource names, leading to incorrect permission evaluations. This allows authenticated users with specific custom roles to gain unauthorized access to subresources, potentially disclosing sensitive information or performing actions they are not permitted to do. Additionally, legitimate users may be denied access to resources.

OpenClaw MS Teams before 2026.5.12 contain an authorization bypass vulnerability where the allowFrom feature binds to mutable display names. Attackers with lower-trust access can perform actions requiring stronger authorization by exploiting the mutable display name binding in the affected feature.

OpenClaw versions before 2026.5.18 contain an authorization bypass vulnerability in skill command dispatch that allows lower-trust callers to execute or persist actions beyond their intended authorization. Attackers can bypass tool policy restrictions through configured input paths to perform unauthorized actions when the affected feature is enabled and reachable.

NATS Server is a high-performance server for NATS.io, the cloud and edge native messaging system. Prior to 2.14.3 and 2.12.12, a client could be registered as the configured no_auth_user through a parser path used when the first client operation was not CONNECT, bypassing user-level connection restrictions such as allowed_connection_types or proxy_required that normal authentication would apply. This issue is fixed in versions 2.14.3 and 2.12.12.

OpenClaw 2026.5.12 before 2026.5.26 contain an incorrect authorization vulnerability in the ClickClack allowFrom feature. When the affected feature is enabled and reachable, a lower-trust caller or configured input path could execute or persist actions beyond the caller's intended authorization, including running non-allowlisted commands.

SurrealDB is a scalable, distributed, collaborative, document-graph database for the realtime web. Prior to 3.1.0, Document::purge_edges in surrealdb/core/src/doc/delete.rs automatically removed graph edge records with permissions disabled through opt.clone().with_perms(false) when a connected node was deleted, bypassing the edge table's PERMISSIONS FOR delete and PERMISSIONS FOR select clauses. This issue is fixed in version 3.1.0.

Umbraco is an ASP.NET content management system (CMS). Starting in version 8.0.0 and prior to versions 8.18.10, 10.8.1, and 12.3.4, users with low privileges (Editor, etc.) are able to access some unintended endpoints. Versions 8.18.10, 10.8.1, and 12.3.4 contain a patch for this issue.

Incorrect Authorization vulnerability in The Wikimedia Foundation Mediawiki - OAuth Extension allows Authentication Bypass.This issue affects Mediawiki - OAuth Extension: from 1.39 through 1.43.

1Password Connect server before 1.2 is missing validation checks, permitting users to create Secrets Automation access tokens that can be used to perform privilege escalation. Malicious users authorized to create Secrets Automation access tokens can create tokens that have access beyond what the user is authorized to access, but limited to the existing authorizations of the Secret Automation the token is created in.

NanoClaw before 2.1.0 contains a privilege escalation vulnerability in the channel-registration approval flow where handleChannelApprovalResponse fails to validate admin privileges over target agent groups. Scoped admins can submit forged or stale connect callback values to wire messaging channels into out-of-scope agent groups, exposing unauthorized groups to unapproved channels and enabling unauthorized observation or control of restricted agent group activity.

An issue was discovered in the AbuseFilter extension for MediaWiki through 1.35.2. It incorrectly executed certain rules related to blocking accounts after account creation. Such rules would allow for user accounts to be created while blocking only the IP address used to create an account (and not the user account itself). Such rules could also be used by a nefarious, unprivileged user to catalog and enumerate any number of IP addresses related to these account creations.

Mattermost versions 10.4.x <= 10.4.2, 10.3.x <= 10.3.3, 9.11.x <= 9.11.8 fail to fail to enforce channel conversion restrictions, which allows members with permission to convert public channels to private ones to also convert private ones to public

n8n-MCP is an MCP server that provides AI assistants access to n8n node documentation, properties, and operations. Prior to 2.57.4, multi-tenant HTTP mode with ENABLE_MULTI_TENANT=true could allow an authenticated tenant to access default-scope workflow_versions backups instead of being confined to the tenant scope, exposing or deleting workflow-version backups from prior single-tenant deployments or migrations. This issue is fixed in version 2.57.4.

Adobe Commerce versions 2.4.7-p3, 2.4.6-p8, 2.4.5-p10, 2.4.4-p11, 2.4.8-beta1 and earlier are affected by an Incorrect Authorization vulnerability that could result in a security feature bypass. A low-privileged attacker could exploit this vulnerability to view or modify select information. Exploitation of this issue does not require user interaction.

OpenClaw before 2026.5.7 contains a sender policy bypass vulnerability in BlueBubbles that allows participants to match allowlist entries through conversation metadata rather than stable sender identity. Attackers can influence conversation-level identifiers to receive agent responses intended for configured senders, potentially bypassing access controls.

In Dataiku DSS before 8.0.6, insufficient access control in the Jupyter notebooks integration allows users (who have coding permissions) to read and overwrite notebooks in projects that they are not authorized to access.

Caddy is an extensible server platform that uses TLS by default. From 2.4.0 until 2.11.3, the authorization layer and the /config traversal layer do not agree on what object the path refers to. In this case, a path authorized for one config object is accepted, but then resolves to a different config object during traversal. This happens because the authorization layer uses string prefix matching and the /config traversal layer parses array indices numerically using strconv.Atoi(). This vulnerability is fixed in 2.11.3.

OpenClaw before 2026.4.20 contains a tool policy bypass vulnerability allowing bundled MCP and LSP tools to circumvent configured tool restrictions. Attackers with local agent access can append restricted tools to the effective tool set after policy filtering, bypassing profile policies, allow/deny lists, owner-only restrictions, sandbox policies, and subagent policies.

Vulnerability in the PeopleSoft Enterprise FIN Cash Management product of Oracle PeopleSoft (component: Cash Management). The supported version that is affected is 9.2. Easily exploitable vulnerability allows low privileged attacker with network access via HTTP to compromise PeopleSoft Enterprise FIN Cash Management. Successful attacks of this vulnerability can result in unauthorized update, insert or delete access to some of PeopleSoft Enterprise FIN Cash Management accessible data as well as unauthorized read access to a subset of PeopleSoft Enterprise FIN Cash Management accessible data. CVSS 3.1 Base Score 5.4 (Confidentiality and Integrity impacts). CVSS Vector: (CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:U/C:L/I:L/A:N).

Vulnerability in the MySQL Server product of Oracle MySQL (component: Server: Security: Privileges). Supported versions that are affected are 8.0.40 and prior, 8.4.3 and prior and 9.1.0 and prior. Easily exploitable vulnerability allows low privileged attacker with network access via multiple protocols to compromise MySQL Server. Successful attacks of this vulnerability can result in unauthorized update, insert or delete access to some of MySQL Server accessible data as well as unauthorized read access to a subset of MySQL Server accessible data. CVSS 3.1 Base Score 5.4 (Confidentiality and Integrity impacts). CVSS Vector: (CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:U/C:L/I:L/A:N).

Vulnerability in the PeopleSoft Enterprise SCM Purchasing product of Oracle PeopleSoft (component: Purchasing). The supported version that is affected is 9.2. Easily exploitable vulnerability allows low privileged attacker with network access via HTTP to compromise PeopleSoft Enterprise SCM Purchasing. Successful attacks of this vulnerability can result in unauthorized update, insert or delete access to some of PeopleSoft Enterprise SCM Purchasing accessible data as well as unauthorized read access to a subset of PeopleSoft Enterprise SCM Purchasing accessible data. CVSS 3.1 Base Score 5.4 (Confidentiality and Integrity impacts). CVSS Vector: (CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:U/C:L/I:L/A:N).

Vulnerability in the PeopleSoft Enterprise FIN eSettlements product of Oracle PeopleSoft (component: eSettlements). The supported version that is affected is 9.2. Easily exploitable vulnerability allows low privileged attacker with network access via HTTP to compromise PeopleSoft Enterprise FIN eSettlements. Successful attacks of this vulnerability can result in unauthorized update, insert or delete access to some of PeopleSoft Enterprise FIN eSettlements accessible data as well as unauthorized read access to a subset of PeopleSoft Enterprise FIN eSettlements accessible data. CVSS 3.1 Base Score 5.4 (Confidentiality and Integrity impacts). CVSS Vector: (CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:U/C:L/I:L/A:N).

IRIS is a web collaborative platform that helps incident responders share technical details during investigations. In versions prior to 2.4.28, users can create alerts for customers that are not assigned to them. This can be abused to falsely attribute fake alerts to customers. In combination with Cross-Site Scripting, this can also be used to exfiltrate alerts from other customers. Version 2.4.28 contains a patch.

OpenClaw before 2026.4.20 contains an improper authorization vulnerability in paired-device pairing management that allows limited-scope sessions to enumerate and act on pairing requests. Attackers with paired-device access can approve or operate on unrelated pending device requests within the same gateway scope.