George Musser wrote a very informative article to the Quanta Magazine, Marolf and Maxfield. When Hawking admitted that he had lost the bet and the information was preserved, he also offered a solution that no one considered kosher. That solution included some new, topologically distinct, configurations that contribute to the path integral and where the black hole is basically absent. The information doesn't stay confined inside the black hole forever because in the most relevant histories, the black hole doesn't really exist, he confusingly argued.
In the most general sense, Marolf and Maxfield do something very similar. But the details are completely different and those in 2020 are far more complex. Marolf and Maxfield postulated lots of null states that determine the difference between the otherwise independent states. Those must exist in every consistent quantum theory of gravity, they claim, and they have linked these null states to replicas, wormholes, and baby universes in rather confusing ways.
A more general provocative statement is that the black hole information puzzle is solved (which means, it is proven that the information surprisingly gets out and roughly how) without details of string theory and, even more provocatively, without considering gravity in a fully quantum way. The new configurations are enough. I think it is right to be supportive of such efforts but at the end, they must be considered inconclusive even if they are tantalizing. It is a big claim that the fully quantum character of gravity is unimportant in such a typical full-blown quantum gravity question. I would prefer a demonstration within a string theory setup (or all of them!) that this simplification is justified; this is something I demand for progress that would look "demonstrably real" to me. Otherwise I do agree with Nick Warner that such "pure QM and GR" arguments have often been circular and there's a risk that it's still the case here. I would even be afraid of the mental "return back to the late 1970s" when these questions were understood very poorly, I think. Some trends simply look negative to me.
Musser has also discussed other papers such as those co-authored by Tom Hartman, a brilliant student whom I could watch at Harvard, and Geoff Penington (who was mentioned more often on this blog in recent years). I feel that the assorted papers are a little bit less directly related than how Musser makes them sound. And again, feel free to disagree but the longer these arguments lead us from constructions that are compatible with string theory, the less likely to be correct they will look to me.