Since that page does not mention Linux, I cannot imagine what relevance it has to grandparent.

As one of the owners of the Chrome Windows sandbox, that level of inaccuracy hurts to read. The correct statement is that Nils and Jon's pwn2own exploit bypassed the sandbox by using a Windows kernel vulnerability that's exposed in every process on the system. You can argue about the extent to which you should be able to rely on the kernel to enforce its guarantees, but the component that failed isn't a point of debate.

Maybe it was just attacked on Windows, the only reference I can find is not super technical and references "zero-days" in the browser itself.