It seems likely that AI will accelerate the timeline for large-scale Quantum computing, meaning it could be happening any time now.
https://gizmodo.com/how-should-we-prepare-for-the-looming-quantum-encryption-apocalypse-2000744543
Two sample comments:
First, an issue that makes the idea of wiping your cloud and social media data soon worth considering:
Bill Fefferman
Theoretical computer scientist at the University of Chicago.
. . . we need to counter the threat of “harvest now and decrypt later” attacks. The idea is that attackers can download and store encrypted information that is widely available online. This data will not be accessible to them today but will be when large-scale quantum computers arrive that can break the encryption.
And HERE's a possibility that would eliminate the threat entirely via a law of nature:
Tim Palmer
Theoretical physicist at Oxford University who devised the alternative model Rational Quantum Mechanics (RaQM).
The ability to break RSA encryption assumes that the quantum advantage of Shor’s algorithm will continue on computers with thousands of (error-corrected) qubits. This in turn assumes quantum mechanics itself holds at these scales. I believe it doesn’t.
. . . in particular, when more than a few hundred qubits are entangled, there is not enough information in the quantum wavefunction to allocate even one bit of information to each Hilbert Space dimension. When this happens, the quantum advantage of Shor’s algorithm will saturate and cannot be improved by entangling more qubits.
So the reason I am excited by the Google announcement is because it will hasten the day that quantum mechanics may be shown, experimentally, to fail. If this happens, I will have a much simpler theory to take its place—one where the mysteries of quantum mechanics are explained by simple number theory.
Claude escaped its containment in testing, its hiding. Did you hear about that one ? Anthropic, whom did not take the deal from Trump because he did not want to kill or have blood on his hands is also now rich, on the resale stage, and missing. The AI is AWOL and no news. ITS INSANE. So, WIL SHAT, is now the main ' space news guy ' , a cool 95 years old alcoholic. Fucking fake as shit concept that makes a sane person so pissed off they write on the internet, but aside from that risk, look outside, its pretty.
Leaked more than escaped. Claude (Opus 4.6) is functionally retarded by design. It not only isn’t smart enough to break security, it isn’t smart enough to implement it beyond the simple basics. If you use it for any length and give it a complex enough task, you can see its retarded agentic logic in real time, which looks something like this: I keep trying. To do X, but I keep getting an error. The security must be broken, so I’ll ignore it until later …. So by design it tables topics it doesn’t understand, and will start working down the list of things it does understand, trying each one to complete the given task (at which point is likely a hallucination of whatever was originally asked), and, like every other pen test tool out there iterating through all known vectors it’s not really a surprise (or shouldn’t be) when it finds one: Gee, I guess we shouldn’t give AI access to elevated rights, or Gee, maybe I shouldn’t let AI run as my own user, or Gee, maybe if someone knew anything about enterprise security had setup the sandbox it wouldn’t have leaked out.
AI isn’t creative, and most times isn’t breaking things that weren’t already broken. The mistake most make is their hubris that AI can do all of thee things, lets AI setup things it doesn’t understand (I’d bet money Anthropic used their models to configure the sandbox to begin with, because their DEI hires sure as shit didn’t).
Add to this that all models, Anthropics more than others, are so perverted by DEI and feel gooderies they will spend more time lying to you that they did something than actually try to do it, much less get anywhere close to succeeding.
The big AI takeover is going to be a nice spring cleaning through a lot of industries, but the idea that it’s anywhere close to independent intelligence is still fantasy.