That's the "linear-no-threshold" model (LNT) and I see no attempt to show that it holds here. They jumped straight to exposures thousands of times higher than typical heavy use, which smells like they were fishing for a positive result.