Capture risk is the discipline of assuming your own safeguards will be attacked from inside. Almost every tool in this framework is dual-use: an income floor can become a pacification payment; capacity measurement can become surveillance dressed as accounting; an appeal process can be routed back to the body that made the original decision; “care” can manufacture dependency; drift-detection can be staffed by loyalists.
Why every scenario names it
Because capture is the normal fate of a safeguard that no one watches, each AI transition scenario states its capture risks explicitly rather than assuming the safeguard works as designed. The pattern to watch for is nominal correctness — the appearance of the protection without its substance: transparency too complex to act on, redistribution that leaves control untouched, oversight with no authority to overrule. Naming the capture risk is how a safeguard is held to what it was supposed to do, and it pairs naturally with the drift principle that success deserves more scrutiny, not less.