Classical OSINT using AI Actual Intelligence

The talk starts from the observation that investigators now have more tools, automation and AI than ever, yet still make familiar errors: accepting the first result, trusting repeated content, and mistaking collected information for intelligence. The speaker argues that when tools are unavailable or wrong, analytical method is what remains, and introduces a framework called ACTUAL. Each letter maps to a step: asking well-defined questions, moving toward primary sources, testing assumptions, weighing context, and linking evidence so every conclusion can be justified with a stated level of confidence.

To show the method in practice, the speaker works through a fictional missing-person case built only from public material. The clues include a library exit time, a bus schedule that had already stopped running, a campus parking structure, a family cabin mentioned in an old newspaper article, mobile coverage maps, and two conflicting sightings. Each clue is checked against the others, weak leads are downgraded, and competing explanations are compared. The route and timing favour a voluntary trip over foul play, but the speaker stresses that a lack of evidence for harm is not proof of safety.

The takeaway is that tools gather material, but the investigator decides what matters. Practitioners should write down their assumptions, look for evidence that would disprove a working theory, and express findings as low, moderate or high confidence rather than certainty.