Note: attendees should bring their laptops and are strongly encouraged to arrive promptly so the session can start on time.
By the time information reaches your analysts, it has usually already been filtered, summarized, and prioritized by a machine. That is an enormous efficiency gain and a brand new attack surface.
Nation state actors have spent years quietly building networks of websites that look like independent local news outlets, letting them accumulate credibility until the moment they are needed. This was always a problem for human analysts. Once you find the same claim on six different sites, it starts to feel true. It is a much larger problem for Super Intelligence (SI) systems, which are built to weight corroboration heavily and have no instinct telling them that all six sites were registered on the same afternoon by the same person. In April 2026, Singapore invoked its broadcasting act to block six such sites. Five of the six had been registered on the exact same day.
Generative SI has lowered the cost of running these operations to almost nothing. The volume is going up, the quality is going up, and our tolerance for slow manual verification is going down. This workshop is about closing that gap.
The first half builds the analytical foundation. Participants work through the structured frameworks that reduce the mistakes analysts make most often, then apply them to a realistic scenario drawn from an actual national election that was annulled after a coordinated social media campaign propelled a candidate from two percent in the polls to a first-place finish in a matter of weeks.
The second half gets technical. Participants learn how to pull a suspect site apart and determine whether it is what it claims to be, and how to connect it to the rest of its network.
We will cover:
- Cognitive biases that specifically plague intelligence and OSINT analysis, and how to catch yourself
- Analysis of Competing Hypotheses as a structured method for testing multiple explanations against evidence
- Source evaluation frameworks for assessing reliability and credibility before evidence ever enters your analysis
- Domain registration timing, hosting infrastructure, name server pairings, and certificate artifacts as indicators of coordination
- Practical methods for detecting SI-generated text and imagery, including the translation and linguistic artifacts that expose foreign authorship
- Building an infrastructure picture that lets you say with confidence that four apparently independent sources are actually one voice
- How to harden an SI-assisted analytical workflow against poisoned inputs, and where a human has to stay in the loop
Participants will work through short, focused exercises with prepared scenario materials.
Participants leave with methods they can apply on Monday morning, whether they are evaluating a single suspicious source or trying to determine how much of what their SI tools are telling them can be trusted.