Direct link to the published paper.
TLDR: They are looking at distant binaries.
(Distant Binaries orbit many astronomical units from each other, Alpha Centauri A+B, are a distant binary system)
The orbital data for these systems shows a lot of variance that should not be there. One issue is there could be a third (or even forth) smaller star (brown dwarf) also present but undetectable that is causing the errors.
The research team tried to eliminate the possibility of these bodies causing the observed errors in the two body data.
They have found there is still something else happening even when this is done.
This has been published in a very respected journal so it will be interesting to see where this leads.