SEC536: Adversarial AI - Penetration Testing AI Systems

BLUF stands for Bottom Line Up Front — a 2–3 sentence opening that states the analyst's principal judgment, confidence level, and most important implication first, so a decision maker who reads nothing else still has the assessment.
They map to fixed numerical ranges: Almost certain (93–99%), Very likely (85–92%), Likely (70–84%), Probable (55–69%), Roughly even (45–54%), Unlikely (15–44%), Very unlikely (3–14%), and Remote (1–2%) — analysts never use 100% or 0% in order to preserve analytical humility.
ACH is a structured eight-step technique, developed by Richards J. Heuer Jr. at the CIA in 1999, that evaluates all plausible hypotheses against the same evidence in parallel and identifies which hypotheses are most inconsistent with that evidence, rather than seeking to confirm a preferred explanation.
The CRAAP test is a five-criteria framework — Currency, Relevance, Authority, Accuracy, and Purpose — used to score the reliability of a source before it's trusted in analysis, with each criterion rated 1–10 for a total out of 50.
It rates source reliability on a letter scale (A, Completely Reliable, through F, Cannot Be Judged) and information credibility on a separate numeric scale (1, Confirmed, through 6, Cannot Be Judged), because the reliability of a source and the credibility of what it reports are assessed independently.
Dator's Four Futures is a scenario-planning framework with four archetypes — Continuation, Collapse, Discipline, and Transformation — used as a completeness check to make sure strategic forecasting doesn't miss discontinuous outcomes by only considering variations on the status quo.