I can buy this argument. But in practice how much does it matter? It's not illegal to say "you may like this terrorist recruitment video", especially when its just a generic recommendation algorithm making that recommendation.
That “generic recommendation algorithm” is anything but. It’s proprietary and serves their interests and completely opaque. Who knows if it’s an algorithm or actually being hand picked?
It's not illegal but they're civilly liable for the harm they cause.
I don't think so. 230 protects them from liability for the video, and 1A protects them from liability for the recommendation. It would be hard to argue that a recommendation isn't speech.