It's still technically just a claim until it's circulated awhile. The biggest journals like Nature are the most notorious for publishing papers that don't replicate.
Being published in Nature basically means that your paper is fine as far as review can see and that it could have a large impact. It retracts all the time.
Yes, because peer review is not very good and not very deep. Retractions almost always come after the paper sees wider circulation and more people with a skeptical eye analyze the paper. Nature itself discusses this:
Peer review evaluates only the paper. It can’t check the underlying correctness of the science, only that the paper clearly communicates what was done and observed.
Science is a process. Publishing a paper is just one step in the middle of that process. Not an endpoint.
Which is exactly what I and the OP have been saying. This is not a discovery but a claim of a discovery, and whether that claim stands up to scrutiny only time will tell.