Simple: you label them as fiction rather than science. The scientific method has experimental feedback built into the loop.

Well then you have to label everything as fiction rather than science because if you don't single out the simplest theory you can't say that general relativity is science and general relativity plus sun turns into cake tomorrow isn't science. And if someone tells you the sun will turn into cake tomorrow and claims that your theory where the sun doesn't turn into a cake tomorrow is fiction and unscientific cause it doesn't make predictions to distinguish it from their theory, you don't have any principled philosophical ground to disagree whith them .

And that's not so clearly bad there because they are obiously wrong, but then you get to complicated topics where everyone has hard to resolve disagreements about which theory should be the default from a big pool of theories that seem to agree whith the evidence.

Doing something clearly outside of empiricism then saying “why doesn’t empiricism cover this” is complete nonsense. What is the ontological explanation for your hypothesis that the sun will turn into cake? Misunderstanding the scientific method is not an issue with the scientific method.