Chifan, Kida, and myself introduced a new class of non-amenable groups denoted by ${\bf NC} \cap {\bf Quot}(\mathcal C_{rss})$ which gives rise to \emph{prime} von Neumann algebras. This means that for every $\G\in {\bf NC} \cap {\bf Quot}(\mathcal C_{rss})$ its group von Neumann algebra $L(\G)$ cannot be decomposed as a tensor product of diffuse von Neumann algebras. The class ${\bf NC} \cap {\bf Quot}(\mathcal C_{rss})$ is fairly large as it contains many natural examples of groups, some intensively studied in various areas of mathematics: all infinite central quotients of pure surface braid groups; all mapping class groups of (punctured) surfaces of genus $0,1,2$; most Torelli groups and Johnson kernels of (punctured) surfaces of genus $0,1,2$; and, all groups hyperbolic relative to finite families of residually finite, exact, infinite, proper subgroups.
In a separate investigation, de Santiago and myself were able to extend the previous techniques that allowed us to eliminate the usage of the {\bf NC} condition and ultimately classify all the possible tensor factorization of the von Neumann algebras of groups that belong solely to ${\bf Quot}(\mathcal C_{rss})$. This provides a far-reaching generalization of the aforementioned primeness results; for instance, we were able to show that if $\Gamma$ is a poly-hyperbolic group, then whenever we have a tensor decomposition $L(\G)\cong P_1\bar\otimes P_2 \bar \otimes \cdots \bar\otimes P_n$ then there exists a product decomposition $\G\cong \G_1\times \G_2 \times \cdots \times \G_n$ with $\G_i \in {\bf Quot}(\mathcal C_{rss})$ and, up to amplifications, we have $L(\G_i)\cong P_i$ for all $i=1,n$.