I'd predict defenses growing, as easy as it is to hack it's just as easy to ask Claude for a security review and fix things plus set up some blacklists.
I've had success with just blocking cloud providers and banning IPs that request unexpected URLs (and reviewing/banning AS numbers if some cloud services provider)
I'm expecting in the future we'll just have lists of bad ISPs that can be applied as pi-hole in addition to better safety practices.
Similarly, if you don't insist on moving so fast you can actually fix bugs and create hardened systems.
And finally, we can also give up on the notion that the Internet needs to be secured using entirely technological means. We don't have that requirement for anything in the physical world or even for other communication channels (e.g. phones, mail). We don't need to make murder literally impossible and we don't need to make computers literally unhackable. Doing so is a choice with a non-zero cost and if the cost increases dramatically we need to re-evaluate the choice and reconsider alternatives.
Some other thoughts:
* Will the Internet become balkanised?
* These smaller networks will require a lot of personal information to even gain access.
* The return of the Sneakernet?