It's so uncharacteristic of the way the IA normally behaves. Whatever special rights a lot of people think they have as a library, a lot of their archiving operations are on somewhat shaky ground. BUT they're a non-profit, don't monetize with advertising, bend over backwards (further than many think they should) to take down anything anyone with a legitimate claim wants taken down even retroactively, and generally don't do anything that makes them a particularly attractive legal target. Yeah, you can argue about them collecting so much content on essentially an opt-out basis. But, at the end of the day, they do provide an opt-out so it's probably not very attractive to sue them when you can get the same result by asking for a takedown.