Can a tag be changed remotely?
No. A tag is written by holding a device against it, at the same couple of centimetres a read needs, so nobody changes a tag they are not standing next to. What can change without anybody touching it is what the address leads to, and that is a feature rather than a risk.
Writing is a physical act
A tag is written the same way it is read: a device is brought against it, powers it, and sends commands. There is no other route in.
No tag has a network connection, an update channel or any means of being reached from elsewhere. It exists only while something is held against it.
So a deployed tag cannot be altered by anybody who is not physically present, which is a stronger guarantee than most systems offer.
The destination is what changes
Almost every tag holds an address rather than content, and an address is a pointer. What it points at can be changed at any time by whoever controls it.
That is why a well built deployment writes an address on a domain it owns and redirects onward: the tags become permanent and the behaviour stays editable.
People who describe updating their tags remotely are describing this. The tags did not change and never will.
It is the single most useful technique in tag deployment and it is the reason the physical immutability above is not a limitation.
What follows for a deployment
Everything written directly onto tags is permanent in practice, because correcting it means revisiting every tag.
So content on the tag should be limited to what will never change, and everything else should be behind an address.
A deployment written with final addresses belonging to somebody else cannot be redirected, retired or corrected, which is how installations become unfixable.
Decide this before the first tag is written. It costs nothing then and is unrecoverable afterwards.
The risk that does exist
Somebody standing at your tag with a phone, rewriting it, which takes seconds and leaves no visible trace.
That is a real and easy attack on any unlocked tag in a public place, and it is the reason to lock public tags rather than any concern about copying.
And somebody covering your tag with their own, which the tag cannot resist at all and which inspection rather than configuration detects.
Both are local, physical and addressed by locking and by looking, not by anything to do with networks.
Frequently asked questions
Can a tag be updated over the internet?
No. A tag has no connection of any kind. Writing requires a device held against it.
How do people update tags then?
They change what the address points at. The tags are untouched, which is why the technique works.
So my deployed tags are safe from remote tampering?
Yes. The realistic attack is somebody standing at the tag rewriting it, which locking prevents.
What should go on the tag itself?
Only what will never change. Everything else belongs behind an address you control.