CVE-2026-93595
MediumCVSS 6.5Exploitation Probability (EPSS)
Low risk21th percentile - higher than 21% of all known CVEs
Summary
A vulnerability in ArcadeDB before 26.9.1 exists in the query_database tool exposed through AI chat endpoints. The tool executes queries without binding the authenticated user to the database context, causing per-type and per-bucket ACL checks to fail. Authenticated users can read data they are explicitly denied at the type level.
Risk Assessment
Attackers can exploit the AI assistant to run queries against restricted types or buckets, exposing sensitive data that would otherwise be rejected through standard query endpoints.
Recommendation
Upgrade ArcadeDB to version 26.9.1 or later, which includes the fix. Review permissions configuration and consider limiting access to AI features where possible.
Other vulnerabilities in ArcadeDB
See all- CVE-2026-93594High
ArcadeDB (com.arcadedb:arcadedb-engine) through 26.8.1 enforces per-type/per-record access-control rules only in LocalBucket, keyed on file id. Query paths reaching data through LSM index files or the TimeSeries engine never invoke that check, so an authenticated user denied readRecord/deleteRecord can still read indexed key values, record IDs, MAX/MIN values, TimeSeries samples, learn record counts, and delete index entries, desynchronizing the index from data. Index and type names are discoverable via SELECT FROM schema:indexes. Affects embedded and server deployments and all transports. Fixed in 26.9.1.
- CVE-2026-65831High
ArcadeDB prior to 26.7.1 allows a reader-role user to submit POST /api/v1/command/{database} with language: js because PolyglotQueryEngine.command, analyze, and registerFunctions do not enforce database-administrator authorization. GraalPolyglotEngine also permits scripts to bypass the allowedPackages whitelist by reflecting from the bound database object to arbitrary host classes.
- CVE-2026-54077High
In ArcadeDB prior to 26.6.1, the IMPORT DATABASE statement did not require administrative privileges and passed its source to the importer without validation. An authenticated user with SQL command access via /api/v1/command or /api/v1/query can supply HTTP/HTTPS destinations to make server-side requests to internal services, or file:// paths to read files accessible to the server process and ingest them as records. The XML importer also permits DTD processing and external entities.
- CVE-2026-75854Critical
ArcadeDB versions before 26.8.1 contain a missing authentication vulnerability in the Redis wire-protocol plugin that allows unauthenticated attackers to read, write, and delete data. Attackers can connect to the Redis port and execute arbitrary commands against any database without credentials.
- CVE-2026-75852Critical
ArcadeDB versions before 26.8.1 fail to enforce SASL authentication on data commands in the MongoDB wire-protocol plugin. Unauthenticated attackers can issue insert, find, update, delete, and create commands against any database by connecting to port 27017 without credentials.
- CVE-2026-75851Critical
ArcadeDB server (com.arcadedb:arcadedb-server) in versions 26.7.3 and earlier fails to propagate the authenticated principal to asynchronous command worker threads. When an HTTP command is submitted with awaitResponse:false, it executes on an async worker whose DatabaseContext has no bound user, causing the scripting authorization gate to become a no-op. A user with only read access to a single database can submit an asynchronous JavaScript command via the /api/v1/command endpoint to run code with unrestricted host access and create a server-wide administrator, escalating to full administrative control. Fixed in 26.8.1.
- CVE-2026-75843Critical
ArcadeDB before 26.8.1 fails to bind the authenticated principal on the gRPC transaction executor thread in beginTransaction, allowing authenticated readers to execute JavaScript commands without scripting authorization checks. Attackers can execute executeCommand with a transaction ID to run unrestricted JavaScript that creates server-wide administrator accounts.
- CVE-2026-67342Critical
ArcadeDB before version 26.7.2 contains an authorization bypass vulnerability in HTTP handlers for time series, batch, Prometheus, and Grafana endpoints. Attackers can access and modify databases without proper permissions by directly calling affected endpoints with arbitrary database parameters.
- CVE-2026-67341Critical
ArcadeDB before version 26.7.2 fails to enforce scripting authorization checks on the SQL DEFINE FUNCTION statement with LANGUAGE js. Attackers with database access can execute arbitrary JavaScript code by submitting DEFINE FUNCTION statements, bypassing security controls intended to restrict scripting to administrators.
- CVE-2026-44221Critical
ArcadeDB versions 21.10.1 through 26.4.1 have two defects: (1) ServerSecurityUser.getDatabaseUser() returns a DB user with an uninitialized fileAccessMap, causing allow-all; (2) ArcadeDBServer.createDatabase() omits factory.setSecurity(...), disabling record-level authorization for new databases. An authenticated user or API token can read, write, and mutate schema on any other database on the same server.
Original NVD description (English source)
ArcadeDB before 26.9.1 contains an access control bypass vulnerability in the query_database tool exposed through the AI chat endpoints. The tool executes queries without binding the authenticated principal to DatabaseContext, causing per-type and per-bucket ACL checks to silently no-op and allowing authenticated users to read data they are explicitly denied at the per-type level. Attackers can prompt the AI assistant to execute queries against restricted types or buckets to retrieve sensitive data that would be rejected through normal query endpoints.
Vulnerability data from NVD (NIST) · CISA KEV · EPSS

