High · published November 8, 2025
⚡ A sneaky DOM XSS vulnerability lurks in versions 0.6.34 and below of Open WebUI! When 'Insert Prompt as Rich Text' is enabled, anyone with permission to create prompts can unleash a payload on unsuspecting users! Think of it like a restaurant that allows customers to customize their orders, but forgets to check if the ingredients are safe. If someone slips in a secret ingredient that’s actually harmful, it could spoil the meal for everyone who tries it! An attacker could insert malicious scripts into the chat window, potentially compromising other users' sessions. This means that with a simple command, they could execute harmful actions, leading to unauthorized access or data manipulation. It’s a recipe for disaster in your offline AI platform!
Think of it like a restaurant that allows customers to customize their orders, but forgets to check if the ingredients are safe. If someone slips in a secret ingredient that’s actually harmful, it could spoil the meal for everyone who tries it! This vulnerability occurs because the prompt body is assigned to the DOM sink .innerHTML without any sanitization. When users run commands to insert these prompts, they can unwittingly execute the malicious payload.
An attacker could insert malicious scripts into the chat window, potentially compromising other users' sessions. This means that with a simple command, they could execute harmful actions, leading to unauthorized access or data manipulation. It’s a recipe for disaster in your offline AI platform! Upgrade Open WebUI to version 0.6.35 immediately to patch this vulnerability. Additionally, review user permissions to ensure that only trusted individuals can create prompts. Implement content sanitization practices for added security. You've got this! By taking these steps, you'll be a security hero in no time! 🛡️