This is much harder than it sounds. Most "Personal Firewalls" are failing completely.
It is very easy to circumvent any filtering attempt by not sending directly, but making other applications to send. And if there is connectivity, there are applications which can send, like the web browser or your mail program.
People call that "leaks", and testing programs "leak tests". I wrote two. The first did cost me ten minutes of work, and any "Personal Firewall" was fooled at this time¹, then they patched (it's an unfair game - it is much easier for the attacker to chose the next available option to send, while the "Personal Firewall" programmers have to spend months of development time to prevent that from happening - and they have to destroy functionality of the operating system to get that to work).
After "Zone Alarm" was ready, and had patched, I spend just another half an hour while dinner on a Saturday evening with my laptop², and again every "Personal Firewall" failed.
I stopped that, because I think, problem was showed.
Trying to prevent applications from sending, which you're running on your system, is a b0rken concept anyways.
If you have code running on your system, and this code manages it to gain administrator rights, you lose³.
Usually, people are working as administrator on Microsoft Windows, so there is nothing to do for an attacker. The clever attacker is running code in kernel space then, ignoring any "Personal Firewall".
If people are careful enough to not work as administrator, then there are hundreds of tricks to gain administrator rights on a Windows box. Usually, it's enough to install a printer driver? or use the scheduling service.
But even if all that would work, trying to prevent applications from sending, which are running on your system, is a b0rken concept anyways.
This is, because deciding which communication should be prevented from happening and which not is not a computable problem. If you're preventing an application from "phoning home" to search for updates or for information about new security holes, you're lowering security instead of elevating it.
And because the "Personal Firewall" cannot decide, it is asking the only person, who should not be asked at all, the person who should be secured and not at all be responsible for security:
They're asking the user.
This makes the concept absurd, even if it would work.
Yours, VB.
¹