A field manual for security leaders who already did the assessment and now have to run the thing.

Most books in this space explain what AI risk is. This one assumes you know, and hands you the instruments: a matrix you fill in, a runbook you rehearse, a register you count, a scorecard you present. Twenty-two of them, as worksheets. It's closer to a set of tools than a set of ideas.

If I had to place it on a shelf next to something, it would sit near a well-run operations manual rather than near a strategy book. The prose is direct, the chapters are structurally identical so you can jump to whichever domain is on fire, and it ends every section with an owner and a date rather than a conclusion.

The one thing that makes it more than a toolkit is the argument running underneath: that human-speed governance is on a clock, and everything you build now is scaffolding for a handoff you haven't made yet. That's the part readers will argue with, and it's the part that will still matter in five years.