How do you manage display sharing using it? Another device? I have a bunch of refurb servers, which I'd like to keep access to not waiting for an engineer to visit it. The budget is tight so proper KVM is not an option.

I use AirPods with my MacBook all the time and no one is ever complained. Does Apple have a secret sauce?

Yes, see my comment about their proprietary codec that is exclusive to Apple platforms only. Won't work on Android. Won't work on Linux. Won't work on Windows. Only works with Apple headphones with Apple phones/laptops.

The outcome is not the same. Regardless of how many will be deported, this will normalize the modus operandi. Been there. Aesthetic is crucial if you don't want it to be turned on you some day.

They are not receiving any benefits. They are not legal literally. Work for cash. No safety net except peers who are often abusive. There are lot of this in EU, just not this visible.

In EU you have the right to use bundled software as long as you own the appliance. Not sure this is true for US.

How does that work? What if the company licenses technology from company A to build product B, but the license is only good for 2 years? What happens 2 years after you buy product B?

Also doesn't Apple, Google, & other remove features from people's smartphones after release all the time in the EU?

This is not legal to sell a finished product which has a license time bomb, I suppose.

Google and Apple can change the future set but they do not brick the device which was discussed and it works as advertised at the moment of purchase.

What is the "future set" ?

Typo. "Feature set", referring to your comment. Functions of software. Quoted below, source is widely available.

“The notion of goods with digital elements should refer to goods that incorporate or are inter-connected with digital content or a digital service in such a way that the absence of that digital content or digital service would prevent the goods from performing their functions.” — Recital 14, Directive (EU) 2019/771.

It can be actually hidden only paired with an HTTPS server, since CT new certificate logs entries are pretty strong indicator that a host deserves to be "visited".

This. The former core user base would prefer the definitive history destruction to a reliable backup. But this audience doesn't grow this fast anymore.

It's changes the data flow. Transit connections are magnitudes expensive than local exchanges, and you can even connect to neighborhood country exchanges on lower prices than serve all TikTok through Ams/Fra. Since VPN is encrypted you can't reroute its content by your rules.

Also mentioned here, larger corps have local caches which unloads transit significantly. Google does this for YouTube everywhere.

You have just described a singularity point for this line of business. Which could happen. Or not.

I wouldn't describe it as a singularity point. I don't mean that they'll get models to design better model architectures, or come up with feature improvements for the inference/training host frameworks, etc.

Instead, I mean that these later-generation models will be able to be fine-tuned to do things like e.g. recognizing and discretizing "feature circuits" out of the larger model NN into algorithms, such that humans can then simplify these algorithms (representing the fuzzy / incomplete understanding a model learned of a regular digital-logic algorithm) into regular code; expose this code as primitives/intrinsics the inference kernel has access to (e.g. by having output vectors where every odd position represents a primitive operation to be applied before the next attention pass, and every even position represents a parameter for the preceding operation to take); cut out the original circuits recognized by the discretization model, substituting simple layer passthrough with calls to these operations; continue training from there, to collect new, higher-level circuits that use these operations; extract + burn in + reference those; and so on; and then, after some amount of this, go back and re-train the model from the beginning with all these gained operations already being available from the start, "for effect."

Note that human ingenuity is still required at several places in this loop; you can't make a model do this kind of recursive accelerator derivation to itself without any cross-checking, and still expect to get a good result out the other end. (You could, if you could take the accumulated intuition and experience of an ISA designer that guides them to pick the set of CISC instructions to actually increase FLOPS-per-watt rather than just "pushing food around on the plate" — but long explanations or arguments about ISA design, aren't the type of thing that makes it onto the public Internet; and even if they did, there just aren't enough ISAs that have ever been designed for a brute-force learner like an LLM to actually learn any lessons from such discussions. You'd need a type of agent that can make good inferences from far less training data — which is, for now, a human.)