CVE-2026-21836 Overview
CVE-2026-21836 is a Broken Access Control vulnerability [CWE-862] affecting the HCL DominoIQ Retrieval-Augmented Generation (RAG) feature. Under certain circumstances, document-level access restrictions are ignored when the system determines which data to return from an AI query. An authenticated attacker can leverage this flaw to view sensitive data that should otherwise be restricted by document-level access controls.
The vulnerability impacts confidentiality but does not affect integrity or availability. Exploitation requires network access and low-level privileges, with no user interaction needed.
Critical Impact
Authenticated users can retrieve sensitive documents through AI queries that bypass document-level access restrictions in HCL DominoIQ RAG.
Affected Products
- HCL DominoIQ RAG feature
- Refer to the HCL Software Knowledge Base Article for specific affected versions
- HCL Domino deployments leveraging DominoIQ AI capabilities
Discovery Timeline
- 2026-05-20 - CVE CVE-2026-21836 published to NVD
- 2026-05-20 - Last updated in NVD database
Technical Details for CVE-2026-21836
Vulnerability Analysis
The vulnerability resides in the DominoIQ RAG feature, which augments AI query responses by retrieving relevant documents from a Domino data store. RAG systems typically index documents and return the most relevant content snippets to the language model for response generation.
In HCL DominoIQ, the retrieval layer fails to consistently enforce Domino's document-level access control lists (ACLs) when selecting source documents for an AI query. As a result, content from documents the requesting user is not authorized to read can be surfaced in AI-generated responses.
The impact is restricted to confidentiality. An attacker cannot modify data or impair service availability through this flaw. However, sensitive documents indexed for RAG, including internal communications, HR records, or business-confidential material, may be exposed.
Root Cause
The root cause is a missing authorization check [CWE-862] in the RAG retrieval pipeline. Document-level access restrictions configured in Domino are not consulted during the AI query data selection step. The retrieval logic returns content based on relevance scoring without verifying that the requesting authenticated user has read access to the source document.
Attack Vector
Exploitation requires an authenticated user account on the target Domino environment. The attacker submits crafted natural-language queries to the DominoIQ AI interface, designed to coax the RAG layer into surfacing restricted content. Because the access check is missing rather than weak, no privilege escalation or credential theft is required. The vulnerability manifests over the network through the standard DominoIQ query interface. Refer to the HCL Software Knowledge Base Article for vendor-confirmed technical details.
Detection Methods for CVE-2026-21836
Indicators of Compromise
- Unusual volume of DominoIQ AI queries from a single authenticated user account
- AI query patterns referencing keywords, project names, or document identifiers the user has no business need to access
- Audit log entries showing RAG retrieval of documents inconsistent with the user's role or department
Detection Strategies
- Correlate DominoIQ query logs with Domino ACL membership to identify responses that include content from documents the querying user cannot directly open
- Baseline normal AI query topics per user or role and alert on significant deviations
- Review RAG indexing scope to identify which sensitive document libraries are exposed to AI queries
Monitoring Recommendations
- Enable verbose logging for DominoIQ RAG retrievals, capturing source document references for each AI response
- Forward Domino and DominoIQ logs to a centralized SIEM for retention and cross-source correlation
- Monitor for spikes in AI query activity, especially from accounts with low historical usage
How to Mitigate CVE-2026-21836
Immediate Actions Required
- Apply the vendor remediation documented in the HCL Software Knowledge Base Article
- Audit which document libraries and databases are indexed by DominoIQ RAG and remove highly sensitive content from the AI index until patched
- Review DominoIQ usage logs since the feature was enabled to identify potential prior exposure
Patch Information
HCL has published remediation guidance in the HCL Software Knowledge Base Article. Administrators should consult the advisory for fixed versions and upgrade procedures specific to their DominoIQ deployment.
Workarounds
- Restrict DominoIQ RAG indexing scope to document collections that are intended to be broadly readable
- Limit DominoIQ access to a smaller group of trusted users until the patch is applied
- Temporarily disable the DominoIQ RAG feature in environments where sensitive documents are indexed and patching is not yet feasible
Disclaimer: This content was generated using AI. While we strive for accuracy, please verify critical information with official sources.

