...the quantum kind and the PR kind. Assuming the sensor works as advertised, the magnetic signature of a Black Hawk's own avionics and rotor blade static would swamp a human heartbeat's magnetic signature by many, many orders of magnitude. Unless the CIA has figured out how to build a helicopter out of wood and plastic, the most believable version of what happened is something like this:

- intelligence narrowed the search area first
- manned aircraft or drones swept the region using standard sensors.
- AI fused and filtered these multiple sensor channels, narrowing the search area further.
- quantum magnetometry contributed only at closer range. A quantum magnetometer-equipped drone swoops in to get a closer look at an object of interest for signs-of-life, and by closer in, I mean less than a meter.

There's a real technology here, NV-diamond magnetometers, but the article is swapping out physics for PR vibes. NV-diamond magnetometry is absolutely legit. MIT, NIST, and others have shown you can detect biomagnetic signals like a heartbeat. No argument there. The catch is range, and the range problem isn't solvable by engineering, because the problem is physics and simple math.

Magnetic fields fall off as the cube of the distance from the source. That's a brutal curve. If you've got a picoTesla-level signal (typical of human neural anatomy at centimeter distances) by the time you're at a meter you're down in the femtoTesla range, and much past that, you now have to separate the signal from the Earth's magnetic field, the magnetic fields of nearby geological formations, and your own sensor platform noise. A tricorder does this all the time on Star Trek, yeah, but current technology? Nope.

Getting a clean lock at ~1 meter on a picoTesla field takes on the order of a minute or so of integration. Push that to a few meters and you're talking hours. Push it to anything that looks like "search from altitude" and the integration time goes from hours to geological epochs. AI helps, but it helps by integrating over time and rejecting noise. It doesn't repeal the inverse-cube law. So the most plausible reading of the article isn't "we found a heartbeat across the desert," it's "we used AI and quantum magnetometry for close-range confirmation of signs-of-life."

Fwiw, It seems pretty clear that the Trump administration's marketing department turned this sidebar on a successful SAR op into a gee-whiz story to impress the local yokels. Quantum sensors are real, but long-range heartbeat detection is still science fiction.

I know I do.

Chips are tested after manufacture. A large number of them have tiny flaws in one or more areas, but work perfectly otherwise. The ones that are flawless go in the highest-end models, the others have the flawed area disabled and go in the lower-end models. Some are just rejected outright.

I think you have the right framing here: the tech is real, the range claim is where things get sketchy. For something like a cardiac signal, you're dealing with ~pT at centimeter scale, and then you take a 1/r^3 hit. By 1 meter you're already down in the few fT range, which is at or below the noise floor of anything you'd be able to loft on a drone, especially outside a shielded lab. At that point detection becomes an exercise in integration for your pet AI.

- sub-meter: plausible with enough dwell time
- 1 meter: maybe, with tens of seconds to minutes
- several meters: hours, assuming ideal conditions
- dozens of meters: not happening on any realistic timescale

AI can absolutely help tease out periodic structure from noise and improve confidence over time, but it's working with whatever signal is physically present. It can't conjure amplitude that isn't there. As the article points out, "it took some time" because they were accumulating weak signals via multiple sensor platforms. I'd ignore the PR blurb from the government and interpret this as sensor fusion with quantum magnetometry in the stack, not stand-off heartbeat detection. The former is believable; the latter runs straight into the inverse-cube wall.

Just a couple weeks ago, I replaced the battery in my 6-year-old Lemur Pro. Not very hard, and now it's great at holding a charge again.

Yes, getting this thing in 2020 cost me 2-3 times as much as today's new Macbook Neo, but I needed a machine I could rely on, that wasn't designed as though I'm the manufacturer's adversary.

I'm sure the three people who bought the vision pro will appreciate it.

Yeah, I read through those... and found that while it described a vulnerability, it was still light on actual exploit details.

Did they compromise the inward facing web interface, or an outward web interface? Did they do it through social engineering, or through malware running on devices on the internal network? Was the malware persistent or was it a drive-by instance running a portscanner in a browser instance?

Basically, the question I have is - would flashing say, openWRT on these devices been enough to prevent network intrusion, or were they already inside the gates to begin with?

Could the point of the brag be that it's newer hardware? My understanding is that successive generations of earthling chips are more vulnerable to malfunction from cosmic rays, etc due to their much higher density.

But I've totally not kept up. Is this still a problem?

Since pertinent information was withheld (that it didn't know), then by your own post you acknowledge it was a lie of omission.

The stupidity of people these days is truly beyond belief. And, yes, get the f off my lawn.

We learned back in the 80s that trying to get a neural net to emphasise what you want is actually very difficult. What it will tend to emphasise are the assumptions that underly the test data, and that's usually a completely different sort of fiction.

The linked articles are remarkably light on details of how the routers were compromised. Were they breached from the internet side due to backdoors or poorly implemented services? Was it some sort of configuration default for remote administration that was just bulk abused? Or were the routers compromised from inside the network by malware running locally on machines, or on malware compromised pages? Was it due to remote code execution or was it due to default admin credentials or easily guessable passwords?

Kind of hard to defend against a threat if they won't tell you how the deed was done.

But was that figure provided by AI?

Even if not, we all know that 793% of all statistics are invented.

If something is inaccurately presented as being the truth, then it is a lie of omission because it is dishonest about the fact that the information isn't actually known.

Gemini is exceptionally bad, as LLMs go. I really have no idea why it is so dreadful, even compared to other LLMs. It isn't context window. and it doesn't seem to be training material either.