1,481 karma · joined February 1, 2023
There's one exception, if there's a vulnerability in the hard-coded ROM or the silicon itself. One was found not long ago and enabled newer iPhones to be jailbroken. It lacks persistence and requires lightning port access, but it can't be patched.
A no-click remote ACE at the unpatchable physical level would be the mother of all zero days, and too valuable to use for all but the highest of targets.
Not really true. There are things worse than death and every AI researcher who is width their salt knows this.
Human level intelligence already inflicts not only great amounts of pain but drawn out, unusual, and unnatural durations of pain. So we know artificially enhanced suffering is a risk of intelligence.
How is that unique to America? What other developed nation doesn't have this rule?
It's about as easy as it would be for an ISP to inject code into an HTTPS page.
The only reason anything works is because Google attestation servers still return a green light for evaluationType=BASIC. Once old devices become rare enough they'll only return a positive attestation for evaluationType=HARDWARE_BACKED.
Go find try and find a single instance of anyone achieving HARDWARE_BACKED with less than a fully stock device.
They are none. No amount of Magisk magic will make it work because it's all taken out of software's hands. Bypasses at that point look like electron microscopes and micro-electronics cleanrooms.
OpenAI has been cutting deals left and right with newspapers, publishers, and other organizations to train on vast high quality private datasets that simply are not available to open source development. Looking at how datasets are at least as important as volume of compute, we can expect the gap between OAI and open source models to continue to widen.
None of this addresses the gigantic (possibly 1:1000) advantage services with constant workload have when it comes to memory bottleneck efficacy.
You know this and I know this.