Today, the main danger for hacking comes from artificial intelligence. And it is no longer theory, it is practice. There are already break-ins based on this technology. Keys and passwords are being compromised again. It was reported this month that more than one billion accounts could be hacked using artificial intelligence. Why? Because there's something to steal...
... Now, back to the topic. Public key cryptography uses randomness generated from sources to generate the private key. Since the generation methods involves large amount of entropy, if given a public key, how would you use AI to tackle the algorithm and solve for the appropriate private key?
-----------------------------
Now back to the subject of the question you asked.
How can we use artificial intelligence to solve the problem of finding a private key if we know the public key?
I could be wrong, but the principle of artificial intelligence is algorithms, it is a program with self-development.
If there was an algorithm to find the private key through the public key, there would not be the cryptography we use. Makes sense?
It makes sense, except for cryptography built on elliptic curves, for the reasons described in the last post.
So, the solution algorithm known to us (not to humans in general) does not exist.
Then, I would use artificial intelligence in another way - I would break the whole amount of computation into billions of components (into groups of large sets of numbers to check), and in a mode of covertly forcing computation on all remote network computers, make them work on the problem in secret. This is similar to the process of covert mining of a cryptocurrency, the task being distributed to all systems available for such a covert attack. Then all that remains is to hope for a result in polynomial time. Naturally, I would apply all known algorithms that reduce calculations when solving discrete logarithm or factorization problems of large numbers.
As for the human social graph and guessing, artificial intelligence will help with passwords if they are not random, but will not help at all with keys, with the pair of public and private key, which are generated without taking into account the peculiarities of the person's personality.
And of course, the best and most effective way to get the public key with artificial intelligence is banal phishing, theft, covert espionage, Trojan horse program and other nasties, with which the artificial intelligence will be loaded in the first place.
In that sense, it's interesting to have a discussion, will our security increase or decrease in the age of artificial intelligence?
It's not as simple a question as it seems at first glance...