Open WebUI is a self-hosted artificial intelligence platform designed to operate entirely offline. Prior to 0.9.6, Open WebUI patched SVG XSS in user profile images and webhook profile images but forgot to apply the same fix to model profile images. The ModelMeta class has no validate_profile_image_url field validator, and the model image serving endpoint has no MIME allowlist or nosniff header. Any authenticated user with workspace.models permission (enabled by default) can store a data:image/svg+xml;base64,... payload in a model's profile image and achieve full account takeover of anyone who navigates to the image URL. This vulnerability is fixed in 0.9.6.
Open WebUI is a self-hosted artificial intelligence platform designed to operate entirely offline. Prior to 0.9.6, a path traversal vulnerability exists in open-webui's cache file serving endpoint that allows any authenticated user to read files from sibling directories outside the intended cache directory, by exploiting an incomplete startswith containment check that lacks a trailing path separator. The root cause is that serve_cache_file() in open_webui/main.py validates the resolved path with file_path.startswith(os.path.abspath(CACHE_DIR)) — without appending os.sep. This allows any path resolving to a sibling directory whose name begins with cache (e.g. cache_sibling, cache_backup, cached_models) to pass validation. This vulnerability is fixed in 0.9.6.
Open WebUI is a self-hosted artificial intelligence platform designed to operate entirely offline. Prior to 0.9.6, Open WebUI's prompt version-history endpoints authorize the prompt_id in the URL but then act on caller-supplied history IDs without verifying that the history row belongs to that prompt (history_entry.prompt_id == prompt.id). This affects /api/v1/prompts/id/{prompt_id}/history/diff, /api/v1/prompts/id/{prompt_id}/update/version, and /api/v1/prompts/id/{prompt_id}/history/{history_id}. An authenticated user with access to any prompt they control, plus a victim prompt_history.id, can read or delete another user's private prompt history. This vulnerability is fixed in 0.9.6.
Open WebUI is a self-hosted artificial intelligence platform designed to operate entirely offline. Prior to 0.9.6, Open WebUI has a Broken Object Level Authorization (BOLA) vulnerability in the builtin search_knowledge_files tool. When native function calling is enabled and the selected model has no attached knowledge bases, an authenticated user can call search_knowledge_files with an arbitrary knowledge_id. The function then returns file metadata from that knowledge base without checking whether the user has read access. This allows unauthorized enumeration of private or restricted knowledge base files. This vulnerability is fixed in 0.9.6.
Caddy is an extensible server platform that uses TLS by default. Prior to 2.11.4, forward_auth copy_headers deletes the exact client-supplied identity header before copying the trusted value from the auth gateway. But when the request later goes through php_fastcgi, Caddy normalizes HTTP headers into CGI variables by replacing - with _. This lets a client send an underscore alias that survives the forward_auth delete step but becomes the same PHP/FastCGI variable. Result: a remote client can inject or sometimes override identity/group headers trusted by PHP/FastCGI applications behind Caddy. This vulnerability is fixed in 2.11.4.
Open WebUI is a self-hosted artificial intelligence platform designed to operate entirely offline. Prior to 0.9.6, POST /api/v1/calendars/events/{event_id}/update validates that the caller has write access to the calendar the event currently belongs to, but does not validate the destination calendar_id supplied in the request body. The model layer then persists the new calendar_id unconditionally. A regular user-role account can therefore create an event in their own calendar and immediately move it into any other user's calendar whose ID they know — bypassing the authorization check that create_event correctly performs. This vulnerability is fixed in 0.9.6.
Open WebUI is a self-hosted artificial intelligence platform designed to operate entirely offline. Prior to 0.9.6, the chat message listener allows non-same-origin input:prompt and action:submit messages, so an external site can set prompt text and trigger submitPrompt() in an authenticated victim session. I validated this with a cross-origin attacker page that auto-posted messages and caused unauthorized POST /api/v1/chats/new and POST /api/chat/completions requests containing attacker-controlled prompts. This enables cross-site forced actions and model/tool execution under victim privileges without consent. This vulnerability is fixed in 0.9.6.
Open WebUI is a self-hosted artificial intelligence platform designed to operate entirely offline. Prior to 0.9.6, backend/open_webui/utils/oauth.py::_process_picture_url calls validate_url(picture_url) on the initial URL only, then invokes aiohttp.ClientSession.get(picture_url, ...) without allow_redirects=False. aiohttp's default is allow_redirects=True, max_redirects=10; the function does not pass the project's AIOHTTP_CLIENT_ALLOW_REDIRECTS env constant either. An attacker with a valid OAuth IdP identity can therefore submit a public URL that 302-redirects to an internal address and read the internal response body via the attacker's own profile_image_url field. This vulnerability is fixed in 0.9.6.
Deno is a JavaScript, TypeScript, and WebAssembly runtime. Prior to 2.7.14, Deno's permission system enforces filesystem and execution restrictions by comparing the requested path against the path supplied to --deny-read, --deny-write, --deny-run, or --deny-ffi. On macOS, that comparison was done at the raw-byte level while the APFS filesystem treats different Unicode spellings of the same name as the same file. That means a program could reach a denied path by spelling it differently than the deny rule. This vulnerability is fixed in 2.7.14.
Deno is a JavaScript, TypeScript, and WebAssembly runtime. Prior to 2.7.10, Deno's node:child_process implementation provided an escapeShellArg() helper used when callers passed shell: true to spawn / spawnSync / exec and friends. On Windows, the helper failed to quote arguments that contained cmd.exe metacharacters and did not neutralize % (which cmd.exe expands even inside double-quoted strings). An attacker who controlled any portion of an argument passed to such a call could inject arbitrary additional commands into the spawned cmd.exe invocation. This vulnerability is fixed in 2.7.10.