To put it bluntly: if you do an offline transaction, you're prone to fraud.
If you expect the senders account chain to have balance x (because that's your offline record for that) and the sender has sent all funds to a different address after you synced that account chain, you receive money that isn't there - kind of like an invalid cheque.
I have no clue how that would work in practice, because to know a random account in advance, you'd have to sync the whole amount of account chains there is (called block lattice in Nano's case).
With a mobile device that's hardly feasible and without a mobile device I don't see how you'd get in contact with people to make such an offline transaction.

That kind of specialized AI is what I imagine to be a use case for locally run AI, too.
After all you don't want to build processes on an AI, where you have zero control over what happens behind the curtains.
That includes feeding potentially sensitive data back to it as well as being unable to control the training data set, its learning, version numbers, etc.

Wouldn't infrared achieve the same at less risk for the eyes of other people and less power consumption?

WLAN depending in the standard in use runs exactly there.

Funny, that you even know about Nano!
...it's one of the often overlooked projects because it doesn't have a ton of fuck-off-money and instead tries to focus on a solid protocol.

Nano has a lot of interesting attributes, but I fail to see how what you describe would work in practice.
If both parties want to make sure there are no shenenigans at play, they need to know about the most recent state of the respective account chains, which essentially requires them to be online for agreeing on said transaction.
But overall Nano is very fast and efficient by design and only a failure in terms of "gainz for Lambo".

As we're here in a thread about AI I should remark that machine-to-machine-payments - in this case: agent-to-agent-payments - would work pretty well with Nano as currency because of the transaction finality (typically less than 1 second) and the feeless nature of transactions.
If AI agents are looking for the most viable way to transfer tiny amounts of value fast and without fees they might find Nano and use it - who knows...
...and just like Nano showed that efficient ways to create digital money are possible I'm hoping for efficient AI models that are economically and ecologically worthwhile.

Not OP, but I read it as "accepting" that kind of money equals getting bought with it and by the people who put the money in the PAC.

I'm glad you specifically pointed out proof-of-work blockchains.
They're very inefficient (economically and ecologically) by design. In almost all cases this design isn't warranted.
Alas there are other designs and while most of them are rubbish as well, a few ones are doing things quite right.

I hope that just like other schemes than proof-of-work were thought of regarding blockchains, there will be AI models that are way, way more efficient and ideally can be run locally - for those cases that can use AI...

Good, but why not 120%?

You missed the part where that was fully automated.
With E2EE someone at the other end of the communication would've sounded the alarm.

But go on normalizing that authoritatian state.
I'm glad that other people see the problem that lies within such surveillance.

In that case I'd be willing to shit the extra mile without shortcuts.

I think the dead should be allowed to reap the respect they earned in life.
In some cases that means you intentionally got to get infected by dysentery just to be able to almost shit enough on their grave.

How long until you have to watch an ad before you can start going with the car?

About time to return the favour and free the USA from the Nazis...
Waiting any longer is risking the step from concentration camp to mass murdering camp.