Looked into my key ring and found only a few RSA2048 keys (used by old Proton users). Apparently most devs use Ed or RSA4096 to sign today. Even Thunderbird (its OpenPGP is convenience first, security second, in a sense that your sec key is not passphrase-protected) generates at least RSA3072, RSA2048 is not even an option!

Though this news might be a joke, it’s totally possible that RSA2048 (or RSA itself) becomes eventually obsolete. Which doesn’t mean cryptography in general will be broken, of course. There are different kinds of "one-way" problems, like Ed, already widely used, based on elliptic curves.

If a faster factorization algorithm is found (though that may be proved to be impossible after all), it’s essentially great news. Even Gauss said, “the dignity of the science itself seems to require that every possible means be explored for the solution” (of primality test and factorization), meaning “We must try everything to find a better way to factor a big number!” (which also implies “a more effective attack against RSA!”).

Though no one wants broken cryptography, factorization is something number theorists would love to do quickly too, if possible at all.

See also [not directly related]: https://en.wikipedia.org/wiki/Logjam_(computer_security)