I think the solution is a better access control model for operating systems. If applications are sandboxed effectively, and users are asked to opt in to specif

I think the solution is a better access control model for operating systems. If applications are sandboxed effectively, and users are asked to opt in to specific permissions in a granular way, the risk is not so high. To paint an example, as a default an a process should only have access to its own memory, and maybe a sandboxed directory in the filesystem. If the process wants to access the network, or access other directories, it should ask me as the user permission before it can do so. You could then imagine extending this through a federated trust model. I.e. if I as a user don't want to bother with granting fine grained permissions each time, I could give blanket trust to all applications verified by a given source. This could be for instance a 3rd party app store, or something like a package manager, which I trust to vet the applications which it hosts. There are always going to be challenges with security, but the solution cannot be that 3 corporations get to unilaterally decide which code is allowed to be executed.