That would be a vulnerability of a naive, always on implementation. But I'd think that risk would be significantly mitigated if the self-destruct system has to be armed first.

e.g. China invades Taiwan, someone then immediately presses the "self-destruct enable" button, which cannot be disabled for a long period of time. If the invasion succeeds, someone transmits the self-destruct code. If it fails they don't, the machine stays intact, and someone gets to work disarming the self-destruct.

I think it's plausible such a device could be built. Just make it battery powered and tamper-proof fail-deadly except when the battery runs down. Size the battery for two or three times the time Taiwan could be expected to hold out against an invasion.