Can someone hack your phone through a tag?
In principle any code that parses untrusted input can have flaws, and the software reading a tag is no exception. In practice the attack that works is social: a tag shows you an address, you follow it, and everything that follows is you being persuaded rather than the phone being broken.
The actual attack surface
Something on the phone reads bytes from a tag and interprets them as a structured message. That parsing is code, and code that handles untrusted input is where flaws live.
This is a real category and it would be dishonest to say otherwise. It is the same category as parsing an image, a font or a document.
What differs is the scale of the opportunity, and it is the smallest of any input channel a phone has.
Why it is narrow
The attacker has to be within a couple of centimetres of a phone they do not control, or persuade its owner to bring it there.
They get a few hundred bytes to a few kilobytes, once, with no ability to retry adaptively or to observe what happened.
The parsing happens inside the platform's own handling, which is maintained and patched, and what reaches an application is already structured.
Compare that with a web page, which delivers megabytes of adversary controlled content, repeatedly, from anywhere on earth. The economics are not close.
The attack that works
A tag in a public place holding an address that leads to a page asking for something. A sticker on a parking meter. A tag placed over a legitimate one.
Everything harmful in that sequence happens after the tap, in a browser, at the hands of the person holding the phone.
Which means the defence is the one that works everywhere else: look at where you are being sent, and be suspicious of anything asking you to install, sign in or pay.
A tag is a delivery mechanism for a link, in the same way a printed poster is, and the fact that it is invisible is its one advantage over paper.
Reducing what there is
Keep the phone updated, which is where parsing fixes arrive and is the only technical measure that matters here.
Read unfamiliar tags with a tool rather than letting the phone act on them, which converts an automatic action into a decision.
Turn off background reading if you prefer, accepting that you lose the convenience, and the mechanics are covered on the smartphone side.
And do not treat this as a reason for alarm. The phone in your pocket faces larger risks continuously through channels nobody worries about.
Frequently asked questions
Is there any real attack surface?
Yes, in the parsing of tag content, as with any untrusted input. It is the narrowest input channel a phone has.
How would such an attack reach me?
Somebody would need to be within a couple of centimetres of your phone, with one attempt and a few hundred bytes.
What actually happens in practice?
A tag offers an address, you follow it, and the page persuades you to do something. The phone was never broken.
What should I do?
Keep the phone updated, read unfamiliar tags rather than acting on them, and judge the address before following it.