Not so sure about that, my Tesla went through a phase were it would randomly drift over and hug the centerline, then eventually cross over it, wobble around a bit, and lurch suddenly toward the edge -- with the on-screen visualization showing exactly what it was doing. Seems to have been silently fixed a few software updates ago, right around the same time it stopped spamming errors about the left headlight E-Fuse and front radar in the service log. Whatever it was on, I don't want any, but could probably make some good money selling it.

If this is as widespread as it seems, perhaps the manufacturers should fix the OS to do something about it. Banning "anything that links known SDK x, y or z" is just whack-a-mole.

One simple approach that comes to mind with minimal thought (aside from the obvious "why the fsck does a Pac-Man game need network access at all?") is having the app manifest contain a list of domains to which the app intends to connect, and block anything else at the OS level -- perhaps blocking everything by default, then permitting IPs when the app uses the system resolver to resolve a whitelisted name. Then scale the degree of (human and automated) scrutiny applied to the app with the size and breadth of the whitelist. You can still make a web browser app or something that gets unrestricted access, but it gets a lot of eyes on it. Maybe have to provide source for review. For most apps, "you get five connect() calls per minute unless your manifest declares a good reason for more" might be helpful as well.

Perfect? No, but it increases the complexity and expense of automated spamming/DoS/scraping attacks without proportionally increasing the cost of legitimate behavior (even undesirable legitimate behavior like ads), and funnels potential malfeasance through smaller and more observable chokepoints. Get caught serving a million subdomains to route spam traffic via your one whitelisted domain? Congrats, your dev certificate is now revoked and none of your apps work anymore, and anyone who paid for them gets automatic refunds.

Yeah, no one elected you as much of anything, so kindly take your cameras and fuck off.

While I agree that the original comment was made with no evidence citation and is likely bullshit, there's certainly plenty of precedent across the national and political spectrum for politicians doing the exact things that they've advocated imprisoning others for.

Unfortunately, if he ISN'T convicted, it will likely be due to the ineptitude and buffoonery of the present administration, not for lack of trying.

This is, after all, the DOJ that failed to convict the guy who threw a sandwich at a uniformed law enforcement officer, on camera, then said "yeah, I threw it."

What if he had just written 1209766 on a post-it note inside of his phone case, completely without context? What if he then refused to provide the pin, the agent saw that note and decided to try it and it nuked the device? I have a feeling that now that the door is open we're going to see all sorts of interesting cases testing the limits of the Fourth and Fifth Amendments.

Stale ads are easy enough to solve, just engineer the monitor to fail a few months past the warranty period, then the customer buys a new one with new ads. Simple!

And in a world where it's considered acceptable for a $3000 refrigerator to fail after 5 years, after replacing one that ran for 30 years, they might even get away with it.

Yeah, about that...

GrapheneOS, at least, does this on recent Pixel hardware (confirmed as of Pixel 9 Pro, unknown on lower models.)

It's doing a couple different things, controllable by toggles in settings:

1) Disable 2G network connections globally, which may break a few weird old international roaming connections, but also blocks some old shitty IMSI catchers. Mainly this is done because 2G has little to no protections, blindly offering up globally-trackable identifiers to anyone who asks.

2) "Network notifications" gives the user a notification if their device responds to a request for globally-trackable IDs outside of an established security context. It does not block these requests, just notifies the user.

Pre-5G standards allow the carrier -- or someone impersonating it -- to request the IMEI/IMSI data before establishing an encrypted context. Without going too nerdy, the normal procedure is to negotiate the session using an ephemeral ID and keys known to the carrier and SIM, then exchange the true data within an encrypted envelope. The "Stingrays" do not (supposedly) know the shared key, so they cannot negotiate the encrypted session and have to request the data in the clear, and the OS can tell you this happened, given hardware and firmware support.

5G is stronger, in that it defaults to authenticating using a token derived from the true subscriber ID using a public key, the private component of which is (supposedly) only known to the customer's home carrier, so neither roaming networks nor interception devices can know the true ID. Of course, there are backdoor ways to do this using a null key for "legacy support," and this is one of the things that Android will notify of.

It remains to be determined how strong those "supposedly" elements are.

I find myself doing this frequently and not always voluntarily. Haven't even bothered to check whether it's Clownflare, but most likely. ... oreillyauto.com ... "ACCESS DENIED." Ok. advanceautoparts.com... thanks for playing. It's almost like they don't know their competitor is one line higher or lower in the search results.

... and fittingly, gotta turn off Javascript to view the link. Best sort of paywall, though -- the variety you can opt out of with one click.

Or, for only $5k, get your own private guillotine with no waiting at all! And a padded neckrest, and a 8K video monitor. $20k for a lifetime* subscription.

I guess they could have it, after an appropriate time without comms, key up on the ICAO emergency frequency and start broadcasting its intentions. "Thank you for jamming the satellite communications! This satellite will self-destruct in two minutes and 45 seconds." Wouldn't be the first time Russia did something like this. Remains open to dispute whether the other major nuclear power(s) have done so as well.

I don't know if anyone really complains about AI finding the case law faster than paralegals do. But judges have been known to complain about AIs making up citations that don't exist and then lawyers signing off on them without checking the work.

... and the nationwide public/private surveillance panopticon, and the internet "age-verification" laws, and the slurping up of data-broker output by law enforcement, and the very existence of data brokers, and the proposed requirement to show ID to activate a cell phone, and all the other things that limit the ability of law-abiding citizens to be unknown and mind their business.

I sometimes find myself working in northern Wisconsin. There's a huge population of folks up there who will chase the census taker off their land, occasionally with a gun, because they don't like The Guv'mint being in their business. And then they'll put up a dozen Ring cameras around their property, so if that census taker shows up again, they'll know immediately and can give him what for.

Some combination of profound cluelessness and a love for the taste of boot.