How can you implement a remote kill switch without also creating a permanent unblockable remote code execution vulnerability? The only way I see to do it would be to have the chipset monitor all network interfaces for the packet of death, and give the chipset the ability to inject packets for the purpose of authentication before it actually kills the system.