The rise of fragmented, cross-platform extremist ecosystems requires OSINT practitioners to blend traditional tradecraft with AI-accelerated workflows. This presentation demonstrates how human expertise and Generative AI can work together to rapidly surface, triage, and contextualize violent-ideology content originating from any online source.
Using a contemporary use-case, we begin with manual discovery, showing how analysts identify extremist indicators across obscure websites, non-traditional platforms, and transient content hubs. Human insight remains essential for spotting anomalies, cultural nuances, and subtle behavioral cues that automated tools often miss.
The session then introduces an AI-driven, repeatable triage workflow for violent-ideology assessment. Using a structured decision tree, AI can rapidly evaluate content from images, videos, memes, comments, messaging apps, and archived posts. It identifies extremist symbols, slogans, coded language, behavioral signals, intent markers, capability indicators, and patterns of escalation — enabling analysts to prioritize high-risk material without compromising analytical standards - where the human takes the next step.
Finally, we explore hybrid social network analysis, where AI accelerates entity extraction and link-graph generation, while humans validate significance, influence pathways, and real-world implications.
Together, these methods show how GenAI serves as a force multiplier, enhancing both speed and scale while keeping analysts firmly in control. The result is a resilient, analyst-centered workflow for navigating violent-ideology threats in an increasingly complex digital landscape.
