1The Machine That Can Be FooledAI systems create leverage, but they also create a new attack surface: the input itself.Strategy 2Shadow AIUnapproved AI use is a control problem, but also a demand signal: people are finding faster ways to work before governance catches up.Strategy 3Poisoning the WellData poisoning turns trusted inputs into strategic risk, making AI decisions look sound while quietly bending them in the wrong direction.Strategy 4The Insider That Isn't HumanAI agents and automated systems are becoming privileged actors; the strategic question is how much authority to grant before trust breaks.Strategy 5Anatomy of an AI BreachAn AI breach is rarely a rogue model; it is a failure in the trust chain linking data, instructions, tools, permissions, and decisions.Strategy