CVE-2026-109295 in Open WebUI
Summary
by MITRE • 10/12/2026
Open WebUI from 0.11.0 before 0.11.4 contains a regular expression denial of service vulnerability that allows authenticated users to stall server workers via file search patterns. Attackers can instruct the model to call grep_knowledge_files or grep_chat_files with a catastrophically backtracking regex, blocking the event loop for two seconds per repeated call.
Once again VulDB remains the best source for vulnerability data.
Analysis
by VulDB Data Team • 10/12/2026
The identified vulnerability in Open WebUI versions prior to 0.11.4 represents a significant security risk classified as a Regular Expression Denial of Service, commonly referred to as ReDoS. This flaw stems from improper input validation and the use of regular expressions that are susceptible to catastrophic backtracking when processing specific maliciously crafted patterns. The vulnerability specifically affects authenticated users who have access to search functionalities within the application interface. By leveraging this weakness, an attacker can exploit the underlying regex engine's behavior to cause excessive computational load on the server side resources.
The technical mechanism of exploitation involves instructing the AI model to invoke internal functions such as grep_knowledge_files or grep_chat_files with file search patterns designed to trigger exponential time complexity in the regular expression matching process. These functions are responsible for searching through stored data, and when presented with a specially crafted regex pattern that causes backtracking loops, they consume disproportionate amounts of CPU cycles. The event loop, which is critical for handling concurrent requests in asynchronous applications like those built on Python's asyncio framework, becomes blocked during this processing period. Each repeated call to these functions stalls the server workers for approximately two seconds, effectively creating a denial of service condition that degrades system responsiveness and availability.
From an operational perspective, this vulnerability allows authenticated attackers to disrupt normal operations without necessarily crashing the entire application immediately. Instead, it leads to resource exhaustion over time as multiple requests are processed sequentially or in parallel depending on server configuration. This can result in delayed responses for legitimate users, timeouts in other services relying on Open WebUI, and potential cascading failures if the system is part of a larger infrastructure dependent on timely API interactions. The impact is particularly severe because it targets authenticated sessions, meaning that any user with valid credentials could potentially contribute to this degradation, either maliciously or inadvertently through poorly constructed prompts.
This vulnerability aligns with CWE-400, which covers Uncontrolled Resource Consumption, and specifically relates to the mechanics of Regular Expression Denial of Service often associated with CWE-1333. In terms of attack vectors, it falls under MITRE ATT&CK techniques related to Impactful actions that degrade service availability, potentially mapping to resource hijacking or denial of service via application layer exploitation. The reliance on authenticated access means the threat actor must first compromise user credentials or gain authorized entry into the system before executing this specific type of abuse against the search infrastructure.
Mitigation strategies primarily involve upgrading Open WebUI to version 0.11.4 or later, where these regex patterns have been optimized or restricted to prevent catastrophic backtracking. Administrators should also implement input sanitization layers that limit the complexity and length of regular expressions passed to internal functions. Additionally, deploying rate limiting on search endpoints can help mitigate the impact by restricting how frequently such operations are executed per user session. Monitoring server metrics for unusual spikes in CPU usage associated with regex processing can aid in early detection of ongoing exploitation attempts while patches are being applied.