[R] Can machine learning make side-channel attacks even stronger?
I'm a mathematician. Not really into computers and/or hacking. But I wouldn't be surprised if the following is indeed true:
Machine learning can be used to identify activity on your computer from traces recorded in JavaScript that measure CPU instruction throughput over time
But I have a different question. Why would one human being ever try to compromise another human being? By cheating, mistreating, abusing and so on. An irrational and counterproductive behaviour.
More on reddit.comSide channel vs timing attacks : cryptography
Downfall side channel attack
This one has been fun.
AWS and GCP have already come out with advisories that they are in the clear, but some PaaS companies know for running their own hardware and serving customer projects are still outstanding.
More on reddit.comside-channel attacks
In short no, just patience, curiosity, and a bit of passion for figuring things out :)
This book (which is out later in the year) looks like it would be a great place to start without a formal background in electronics or computer science: https://nostarch.com/hardwarehacking - it's cowritten by Colin from NewAE (https://wiki.newae.com/Main_Page) which also has tons of information on using the ChipWhisperer for SCA stuff. If you have no embedded systems knowledge at all, start with Arduino, move up to writing some bare C, and then you'll be in a better place to jump in on this sort of material.
Good luck!
More on reddit.comFor those who don't know here is a neat description of it from the wikipedia:
"In computer security, a side-channel attack is any attack based on information gained from the implementation of a computer system, rather than weaknesses in the implemented algorithm itself (e.g. cryptanalysis and software bugs). Timing information, power consumption, electromagnetic leaks or even sound can provide an extra source of information, which can be exploited."
It looks like a very fascinating branch of studies. I mean break and explore vulnerabilities from radio, electromagnetic leaks, audio, and other things are super groovy.
Also, I am particularly curious about it because I plan to include this as one of the major plot points in a book that I am planing inspired by Gibson, Stephenson, Watts, and Egan. Resumidely my idea is to write about a hacker group operating in the middle of the decade of 2030 composed of transhumanists, occultists, biohackers, and more other unique people. Who decide to migrate from AI-powered spear-phishing and identity theft to other forms of attacks like side-channel vectors. So I am looking to resources to learn more about this type of attack to satisfy my intellectual curiosity, thanks in advance.