It's building out federation and you can already host your own account and control your own data and run your own feeds. The investors don't have control of the company (public benefit corporation).
The only difference between it and Mastodon in terms of scraping is that scraping public data is a bit easier. Nothing about Mastodon makes that scraping difficult, it's just more annoying to do. The company itself is not doing AI BS.
Local NAS, local security cameras, in-house streaming, LAN multiplayer, local torrent-like data sharing (FYI, Windows Update and more uses the local network to share update between computers by default, so it gets downloaded once and then shared internally)
There's already others running other relays. The appview is the most difficult, but in theory indexes on top of the data could be shared too which would reduce the need for each appview to have everything saved in advance
You don't have to deal with being sent links to other people's mail servers' public mail lists, and then figuring out how to get your own mail server to figure out how to join that conversation. Mailto links open in your mail client which already knows what your server is.
Mastodon don't have a mailto: equivalent, pages can't identify themselves in a way your browser recognize as a Mastodon host, your browser won't remember your Mastodon account(s) specifically. And federation sync issues aren't even dealt with here...
That's what bluesky's DID based account identity does. Unless you make that key the sole authority (no key rotation, like nostr) then you need a registry as authority (like bluesky's PLC registry)
Bluesky specifically lets the account hosting server handle your auth, the directory points to where your account server is as in when you enter your handle, every 3rd party service and federated peer can do OAuth seamlessly to your account host. Then you can log into every compatible site with your handle, instead of having to get redirected "home" before you can interact. Your account server's repository keeps records of all your posts and your social network, and you can even migrate seamlessly across hosts.
They're actively working on making it easier to fork the entire platform.
There's already 3rd party account hosts, moderation labelers, feed generators, and alternative implementations like whitewind (blogging system on top of the same account repo & lexicon architecture) were you can use the exact same account. All of these works together independently of which hosts / providers you use.
Relays and microblogging appviews (a fully functional bluesky mirror) are technically possible but more expensive to duplicate but there's work on making that practical too.
You can already bypass PLC by using DID:Web for account identity.
The single most important thing if you want to be able to recover from bluesky going bad is to just back up your account recover signing key, and to keep a recent backup of the repo to preserve your post history and social network (follows, etc).
All the rest can be rebuilt if you have your account hey and backup. The content addressing will make it seamless!
It's open source and designed to literally not be reliant on the company running it. Start a community appview and plc and you can bring your entire account history with you
They also disagree on basic reality and thinks things are good because of Trump when Trump isn't even in power yet and things are exactly the same as yesterday. It's not just policy preference, or even regular politics, it's cult behavior.
Multiple VPN apps on Android allows you to set per-app rules including forwarding to different VPN servers
The Cloudflare app allows setting exclusions
However you can't have 2+ active VPN connections simultaneously on Android without root, so while for example OpenVPN allows you to set multiple VPN profiles with different app exclusions (binding chosen apps to a certain VPN connection) you're forced to pair this with a firewall to keep different groups of apps offline while their VPN profile is inactive, then switch which apps are online by switching VPN profile.
Alternatively - set up a single Wireguard VPN to your own server somewhere (it may be a rented VPS) and then set multiple outgoing VPN connections on that server, and then set forwarding rules based on which domains/IP each app communicates with (beware that this may make a mess with browsers and such if for example a single web page gets split over different VPN sessions due to content being hosted across different servers on the same page)
It's building out federation and you can already host your own account and control your own data and run your own feeds. The investors don't have control of the company (public benefit corporation).
The only difference between it and Mastodon in terms of scraping is that scraping public data is a bit easier. Nothing about Mastodon makes that scraping difficult, it's just more annoying to do. The company itself is not doing AI BS.