Executive brief
The iAI Lab PDF AI application for Android, which is used for managing and interacting with PDF documents, contains a security flaw that allows other malicious apps on the same device to write files into its private storage. By sending a specially crafted file-sharing request, an attacker can bypass Android's security sandbox to modify the app's internal databases, settings, and cached files. This could lead to a total compromise of the app's data, including user preferences and document history, potentially allowing an attacker to manipulate the app's behavior or persistent state.
Technical details
A 'Dirty Stream' path traversal vulnerability exists in the iAI Lab PDF AI App (chatpdf.pro) on Android. The application exports several activity aliases that handle ACTION_SEND and ACTION_VIEW intents with a mimeType of '*/*'. When processing a content:// URI from an incoming intent, the app retrieves the '_display_name' metadata and uses it unsanitized as a filename for local caching via the getExternalCacheDir function. An attacker-controlled ContentProvider can return a '_display_name' containing '../' sequences, allowing the attacker to write arbitrary bytes to any location within the app's UID-protected internal storage (/data/data/chatpdf.pro/) or external storage. This bypasses the Android application sandbox, enabling the modification of MMKV preference stores, SQLite databases, and configuration files. No fix has been reported by the vendor.
Affected products
- iAI Lab PDF AI App 4.21.0, 4.22.0
Timeline
- 2026-06-06: disclosed: Public disclosure of the vulnerability and exploit code.
- 2026-06-06: advisory