ParsCN: A Persian Dataset for Counter-Narrative Generation to Combat Online Hate Speech

Collection of recent research with direct implications for information ecosystem security and platform integrity. Key findings: (1) Risk-controlling recommender systems that rely on aggregate binary feedback are exploitable — a coordinated group as small as ~1% can degrade recommendation quality by up to ~20%. (2) Agentic operationalization of the DISARM framework scales FIMI (Foreign Information Manipulation and Interference) investigations and revealed previously undetected state-linked bot activity, improving analytic throughput but also demonstrating how agentic tooling changes operational