It helps to keep in mind the recent changes in (A) the political situation in some countries including the U.S. and (B) the capabilities of AI. What was computationally expensive 5 years ago, and reserved for high-profile targets, would be pretty easy now. Recent phones can do a lot of AI processing directly on the device, for instance to accumulate interesting audio over time based on key words (in multiple languages) without draining the battery too much. And of course, once that info is uploaded to a data center and aggregated, there’s no telling what modern AI models could do with it.
A lot of mass surveillance is technically possible now. Are “they” doing it? Will “they” be doing it in 2 months? Do “they” respect laws and borders?
… and within 10 minutes of plugging in the dead phone, something will appear on screen to tell you it’s charging. You haven’t pressed the power button, the phone is not turned on yet, but some code is running to drive the screen and do some charge management.
ETA: …and, as Princhester says, to sense the power button itself, which is just an input after all.
Modern microprocessor controlled devices now commonly use a “soft off” power button meaning they are not usually entirely off. The power button does not physically disconnect the device from its power source and the device remains on (even when “off”) at a very low level, sufficient to detect the power button being pressed, to turn back on. If the device were “completely off” in the sense of becoming totally unpowered when the power button were pressed, then as @Mangetout says it would not be able to turn on again as there would be nothing to detect you pressing the power button.
If you removed the battery the situation would be different - in that case the phone would be totally off. However, if a battery is reconnected I suspect that would trigger the low level circuit “listening” to the power button to automatically re-awaken, allowing the power button to work.
If the battery is merely so low the device shuts off, that is because modern battery management systems turn their associated device off before the battery is completely dead. It’s probably still running at some level.
I’ll do some more measurements and get back with ya. One possible cause for the strange results is a gross inaccuracy in the cheap measurement device I am using. I have no idea how accurate it is.
A lithium battery is never completely dead unless it’s dead as in ‘spicy pillow’. The device may have shown the percentage going down to zero, but the battery still has some residual charge - it never fully discharges intentionally or else it would destroy the battery.
You’ll often find that a device with a really depleted battery (like one that has been in the back of some cupboard for three years) will take a while to begin charging when you plug it in - this is because the microcontroller that is still running has set its own watchdog interval to some larger value, so as to further reduce its own power consumption - so it does not immediately notice that something has been plugged in - that only happens when the timer interval expires and the chip wakes up out of hibernation to briefly look at the state of things.
I feel pretty sure you’d get a different set of results if you were to measure the power consumed from the battery when the phone is in the same ‘off’ state - when it’s not plugged in, and it’s turned off, it should be sipping some tiny little bit of juice just to be ready to come back on (and keep the realtime clock working etc).
When it’s in the same ‘off’ state, but plugged in, I would imagine it might not bother with the hibernation cycles and just runs the standby stuff continually.
And, notably, the emergence of mass-market smart phones, with the first iPhone in 2007 (and the current app store paradigm where it’s easy to wirelessly download software directly to the device), a mere 5 or so years after 9/11. Coincidence? Today 90%+ of Americans carry one of these (and Android) devices in their pocket every day. I think the opportunity to surveil (with efficiencies of AI emerging) is too great to ignore.
There is a lower level, non-SoC driven mode we call Dead Battery Recovery that trickle charges a battery up to a couple of percent and then boots the SoC.
In ye olden tymes there would be an LED indicating it was connected and charging. Now it’s just the time it’s charging before the screen comes on with the Sign Of Life.
Goggling, it could be spyware/malware operating the phone while it’s supposedly turned off. But it could also be an old or damaged battery leaking power, or a hardware issue.
I have a similar model that measures USB-C current, and it always claims an extra 300-400mA of current is flowing, even when it should not be. This makes the device useless to measure how much energy a battery can hold, because it will continue to record 300mA flowing into the battery even after the battery has finished charging.
I have an older one that is USB-A, and it works properly.
Except that the hardest part of fixing a vulnerability is [i]also/i] discovering the vulnerability. Once a vulnerability is discovered, it’s a day or so, tops, before it’s fixed.
Adding to what LSLGuy said, there’s also a resource asymmetry. Big tech can presumably afford to use their own AIs to fuzz their own products, but smaller companies who can’t afford as much AI will be subjected to threats by better-resourced adversaries, whether that’s nation states or blackhats with botnets.
Its basically an AI arms race and whoever has the best models and the most hardware will have an advantage.
It would probably also sharpen the divide between, say, Israel’s cyber offensive capabilities and some of their adversaries’. Stuxnet on steroids.
And now that we’ve cut China off from Nvidia, they’re forced to develop their own chips, over which we have no visibility or control. We’ll see if that ends up biting us in the ass as yet another field they’ll soon leapfrog us in… makes TSMC even more important…
Learnt quite a bit from this thread about what constitutes ‘off’ as far as a phone goes, but wouldn’t a portable ‘Faraday cage’ be a surer method to ensure that your phone wasn’t able to broadcast its location or what its camera and microphone might be recording?