Raw vector
CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:U/C:N/I:H/A:LCVSS and EPSS are reproduced from their sources (NVD, FIRST EPSS). Risk Priority is our own derived reading, not an NVD score.
Summary
CVE-2026-28788 is a high-severity Authorization Bypass Through User-Controlled Key (CWE-639) vulnerability in Openwebui Open Webui. Its CVSS base score is 7.1 (High).
Operationally, exploitation aligns with the MITRE ATT&CK technique Exploit Public-Facing Application (T1190); ranked in the top 14% of CVEs by exploit likelihood; it is not currently listed in the CISA KEV catalog; a public proof-of-concept is referenced.
This vulnerability is AI-related — categorised as LLM Application Platforms; in the LLM/Generative AI Risks risk domain.
The strongest mitigations our analysis identified map to AC-24 (Access Control Decisions) and AC-3 (Access Enforcement) — see the control section below for these in your framework.
Deeper analysis AI-assisted summary
Synthesised by an AI model from the NVD description and linked references — a reading aid, not an authoritative source.
CVE-2026-28788 affects Open WebUI, a self-hosted artificial intelligence platform designed for offline operation. In versions prior to 0.8.6, the vulnerability resides in the `POST /api/v1/retrieval/process/files/batch` endpoint, which allows any authenticated user to overwrite the content of any file by its ID without performing an ownership check. This issue, classified under CWE-639 (Authorization Bypass Through User-Controlled Key), carries a CVSS v3.1 base score of 7.1 (AV:N/AC:L/PR:L/UI:N/S:U/C:N/I:H/A:L), indicating high severity due to network accessibility, low attack complexity, and requirements for low privileges.
An attacker with regular authenticated access can exploit this by first obtaining file UUIDs from a shared knowledge base via the `GET /api/v1/knowledge/{id}/files` endpoint, assuming they have read access. They can then escalate privileges from read-only to write by submitting overwrite requests through the vulnerable batch endpoint. The manipulated file content is subsequently served to the large language model (LLM) via Retrieval-Augmented Generation (RAG), enabling the attacker to control the information presented to other users querying the knowledge base.
The official GitHub security advisory (GHSA-jjp7-g2jw-wh3j) confirms that Open WebUI version 0.8.6 addresses the issue by implementing proper ownership checks in the affected endpoint. Security practitioners should upgrade to at least version 0.8.6 and review access controls for shared knowledge bases.
This vulnerability is particularly relevant in AI/ML environments, as it undermines the integrity of RAG pipelines in self-hosted deployments, potentially leading to prompt injection or misinformation propagation without requiring elevated privileges. No public evidence of real-world exploitation has been reported as of the CVE publication on 2026-03-27.
OWASP Top 10 for Web (2025)
EU & UK References
- 🇪🇺 ENISA EUVD: EUVD-2026-16482
Vulnerability Data
Open WebUI is a self-hosted artificial intelligence platform designed to operate entirely offline. Prior to version 0.8.6, any authenticated user can overwrite any file's content by ID through the `POST /api/v1/retrieval/process/files/batch` endpoint. The endpoint performs no ownership check, so a…
more
regular user with read access to a shared knowledge base can obtain file UUIDs via `GET /api/v1/knowledge/{id}/files` and then overwrite those files, escalating from read to write. The overwritten content is served to the LLM via RAG, meaning the attacker controls what the model tells other users. Version 0.8.6 patches the issue.
- CWE(s)
AI Security AnalysisAI
- AI Category
- LLM Application Platforms
- Risk Domain
- LLM/Generative AI Risks
- OWASP Top 10 for LLMs 2025
- None mapped
- Classification Reason
- Matched keywords: artificial intelligence, llm, open webui
Related Threats
MITRE ATT&CK Enterprise Techniques
CVEs Like This One
Affected Assets
Mitigating Controls
Mitigating Controls (NIST 800-53 r5) AI
Enforcing approved authorizations on every access request structurally stops a user-controlled key from reaching another user's data.
Requiring explicit access-control decisions on each request blocks unauthorized key-driven access.
Least-privilege restrictions limit the scope of data reachable even if a key check is bypassed.
Mitigating Controls (NIST CSF 2.0) AI
Derived directly from the weakness types (CWEs) cited in the NVD entry via our AI-authored CWE→CSF cross-walk (authority under review) — links open the control.
Enforcing authorization policy and least privilege directly blocks user-controlled key tampering that bypasses access checks.
Logical access controls prevent unauthorized data access that results from missing authorization checks on object references.
Mitigating Controls (ISO/IEC 27001:2022 Annex A) AI
Derived directly from the weakness types (CWEs) cited in the NVD entry via our AI-authored CWE→ISO cross-walk (authority under review) — links open the control.
Security testing can detect missing authorization checks but does not prevent the weakness in production.
Information access restriction explicitly enforces that users may only retrieve data they are authorized to see, directly addressing user-controlled key bypass.
Access control policy directly requires enforcement of authorization rules that prevent unauthorized access via manipulated keys.
Managing access rights includes ensuring users can only access their own records and not bypass authorization by altering identifiers.
Privileged access rights control restricts what data each user may access, mitigating direct object reference attacks.
Secure development lifecycle includes authorization design but does not itself implement runtime access checks.