"Whelllp, it turns out that quantum primes aren't the same as classical primes, who'da thunk it?"

Shor requires very good qubits to work, with a very low error rate. When someone will have a QC good enough to actually run it, even for small numbers, they will be screaming from the rooftops about it.

Willow has only ~99.7% 2-qubit gate fidelity and poor connectivity. This is many orders of magnitude too low to run Shor's algorithm for a 50-bit number. The improvement has also stalled quite a bit over the last years (in 2019 they had 99.5% fidelities with similar qubit numbers).

Qbits and error rates have both improved by a few orders of magnitude. Google's paper from August is the first demonstration of error correction. We're probably ~5 years away from doing toy examples of Shor's algorithm with error correction, and ~20 from being able to factor numbers as big as a classical computer.