Em Adespoton

How are people running their Macs exposed directly to the Internet?

In order to make this work, a user would have to:

1. Turn on screen sharing
2. Turn off the OS firewall block
3. Set up NAT routing on their router to forward port 5900 to their Mac

Seems to me that anyone who could do this would know it was a really bad idea?

Especially since you can already tunnel screen sharing via Apple ID or FaceTime to connect without leaving the port open.

The machine is definitely Turing-complete, and definitely NOT trying to pretend to be human. It’s AI claim is for a cloud-based feature I don’t use that adjusts the cycle stages over time to maximize cleanliness and minimize wear on the contents based on personal usage patterns and detected content types.

The machine does do a few neat things like check water retention of the contents before starting the load (it weighs the contents, adds a known volume of water, drains without spinning and checks the weight difference and the dirtiness of the drained water). There’s no way I want it feeding all my washing data into a massive online data lake for AI analysis though, so I keep that bit disconnected.

Well that’s OK; we recently discovered that things like passkeys usually aren’t using the TPM anymore anyway. Most stuff stored in there can be routed around via tokens stored on main storage.

Ironically my washing machine has a big AI logo on the front.

So what’s to stop a network from spinning up a subsidiary MVNO to handle their pre-paid plans and only allocate limited slices to them?

Because it seems like with a bit of paperwork, this would do a total end run around these guardrails.

This is only the case if these changes actually cause a dip in sponsor revenue.

Personally, I like the ones that mimic old-school TV and radio, where the product/service placement is woven right into the show, and there’s a direct relationship between the presenter and the advertiser.

The entire ad brokerage ecosystem could vanish tomorrow and I’d be very happy.

It was also built in to all stereo systems long before that. Not that LPs ever had ads, but I could dub to tape off the radio and then fast forward through the ads.

The thing I find interesting here is that it’s essentially a commercialization of what’s been done in army, police, gang and extremism recruitment for the entirety of history; the tools change, but the methods stay the same.

Last year, you’d be correct. This year, bots DO purchase things. They also download and run things. Some marketers have gotten wise and have adjusted SEO and general product advertising to appeal to bots.

It’s worth also saying that “VPN” is a pretty generic term. It includes:

traditional private networks where a node network or endpoint is securely connected to a local hub network, using the Internet as a transport layer (think Cisco VPN)

Ad-hoc private bridges where two endpoints have a private connection over a public network (think Tailscale)

Residential proxy services where companies can use consumer IP addresses to exit to the Internet, looking like a consumer instead of a business (think Hola VPN / Bright Data)

Network bridges that obscure the source and destination addresses (TOR and I2P)

And THEN… the one that gets all the airtime on YouTube (and is mostly what’s discussed in this article): exit proxy services where a consumer can exit to the Internet from an address other than the one assigned by their ISP (Mullvad, Nord, WindScribe, etc.)

The annoying thing is that marketing for public VPN services often leans on features of various of these configurations where the specific implementation being sold is actually quite limited.

Huggingface (or some other benchmarking company) needs to put its servers in a country with real cyberintrusion laws with teeth. Somewhere where an announcement like this would result in prison sentences and the company’s license to operate being revoked in the country until they could demonstrate via third party audits that their operations are acting in a secure and legal manner.

I have agents force fed to me at work. I’ve already got my own honed workflows I’ve refined over the years; now, Outlook and Teams and Various agentic frameworks are spamming me with summaries of stuff I either already know or don’t need to know… except they still get it wrong constantly, so I can’t trust what they think I should prioritize (usually it’s work I’ve already assigned to others).

Claude Code agents are decent, but they need to create discrete work objects that can be properly reviewed, so act like junior coders.

The main issue I have with most agents today is that they try to replace MY work, making it more complicated and time consuming, instead of completing the work I can’t bother to do myself. And in order to do those things effectively, I have to loosen my privacy and security models to the point where they effectively don’t exist. Which I’d never do for another human, so why should I do it for an agent that’s more likely to catastrophically screw things up at record speed?

I don’t think they’re all lying; most are just misinformed or using the term differently than Cory.

It’s like saying “global warming is changing everything.” Technically, not true, but the knock on effects of specific aspects of it recently hit an inflection point that causes it to affect the lives of everyone.

This has also happened in the field of artificial intelligence; LLM chatbots are only the visible mushroom fruits of the vast mycelium network that has been silently growing underground for years.

They’re actually usually optimized to get AI to serve you the download link… or in the case of agentic browsers, just find, download and run it for you.

They can’t have it both ways:

it takes a picture of a car. Most people go, “this makes a lot of sense. I want my kids to be safe.”

If it’s just taking pictures of cars, it’s doing nothing to keep kids safe. I take pictures of cars. Does that somehow make kids safe?

However, if the argument is: “the system knows where all cars in the area are and have been, and so can keep kids safe…” well, that STILL doesn’t keep kids safe, just allows some understanding of what caused the problem after the fact. But it also lets those with access track where every vehicle is. And then tie in the OTHER cameras the system has access to, and you’re no longer tracking cars, you’re tracking people (and dogs, apparently).

Instead, they abandoned hardware.

My Minolta Dimage was pretty decent in it’s time; I used it at a wedding and got better photos than the professional photographer using film. It was all about understanding the device’s strengths and limitations. The Dimage was great because it had a solid burst mode, so I’d take 1,000 photos where the professional took 30.

Of course, I still eventually moved on from that camera to a decent DSLR camera, which I still use when I go somewhere with the intent to take composed photos.

But the old rule still bolds true: the best camera to use is the one you have on you when the photo presents itself. And everyone tends to have their phone handy, even if they have limited control over focal depth and aperture and try to make up for it in post.