D-Scan

When Context Bites: Detecting RAG Poisoning via Document-Level Attention Collapse D-SCAN is an analysis framework for detecting document poisoning attacks in RAG Retrieval-Augmented Generation systems. It collects LLM internal states during generation token probabilities and attention weights,...

Webfroot Shoutbox 2.32 Remote Command Execution Vulnerability

No description provided by source. source: http://www.securityfocus.com/bid/7746/info Shoutbox is prone to an issue that may result in the execution of attacker-supplied code. The vulnerability exists due to insufficient sanitization of the 'conf' URI parameter. An attacker can exploit this...