CVE-2026-19754
HighCVSS 8.6Exploitation Probability (EPSS)
Low risk32th percentile - higher than 32% of all known CVEs
Summary
Baserow versions 2.2.0 through 2.3.3 contain an SQL injection vulnerability in the index() formula function. A low-privileged authenticated user who can create or modify Formula fields can supply an undocumented fourth argument treated as a SQL template and interpolated directly into a PostgreSQL expression. The generated SQL runs with the Baserow PostgreSQL role permissions, allowing data outside the user's workspace permissions to be read or modified.
Risk Assessment
Low privileges are sufficient to bypass access controls and access or modify other users' data. This can lead to data leakage and database integrity violations.
Recommendation
Upgrade Baserow to a version later than 2.3.3 that contains the fix. Until then, restrict the ability to create and modify Formula fields to trusted users.
Other vulnerabilities in Baserow
See all- CVE-2026-113359Medium
Baserow before 2.3.0 contains a missing authorization vulnerability that allows any authenticated user to read field values from other workspaces via the unique_row_values API endpoint. Attackers can enumerate sequential field IDs to retrieve up to 100 distinct cell values per field from tables in workspaces they do not belong to.
- CVE-2026-81335High
Baserow dispatches an Application Builder data source without acting on the result of its permission check. The dispatch and record-name views are declared with a permission class that admits any caller, and the permission check does not halt execution. An unauthenticated request can receive rows and fields from the data source, and identifiers are small integers that can be enumerated. Version 2.3.1 passes raise_exception to the same call.
- CVE-2026-76837Medium
Baserow interpolates a user's display name into the rich-text mention markup without HTML encoding. PATCH /api/user/account/ stores the first_name value verbatim, and the mention renderer in web-frontend/modules/core/editor/mention.js builds its element with a template literal that places the name into a data-label attribute and the element body unescaped. A name containing a double quote closes that attribute and the opening tag, so following markup lands in the element content. The result is rendered through v-html by the rich-text field components, so any rich-text cell mentioning the account executes the stored script for every workspace member who views the table, without a click. Any workspace member, at the lowest permission level, can set the name and reach every other member of the workspace, and the payload persists until the name is changed or the mentioning row is removed. Version 2.3.0 escapes the value before interpolation.
- CVE-2026-18817Low
Baserow up to version 2.3.2 has an issue in the BaserowImpersonateAuthTokenSerializer function within the Inactive Non-Staff User Handler component, leading to improper authorization. Remote exploitation is possible but requires high complexity. The maintainer suggests it may be more of a bug than a vulnerability, but recommends upgrading to 2.3.3.
- CVE-2026-18816Medium
Baserow up to version 2.3.2 has a vulnerability in the verify function in views.py of the 2FA Verify Endpoint component, leading to improper authentication. The attack can be launched remotely but has high complexity. The issue is fixed in version 2.3.3.
Original NVD description (English source)
Baserow versions 2.2.0 through 2.3.3 contain an SQL injection vulnerability in the index() formula function. A low-privileged authenticated user who can create or modify Formula fields can provide an undocumented fourth argument that is treated as a SQL template and interpolated directly into a PostgreSQL expression. The generated SQL executes through Baserow's database connection with the Baserow PostgreSQL role permissions, allowing data outside the user's workspace permissions to be read or modified.
Vulnerability data from NVD (NIST) · CISA KEV · EPSS