It’s the linear model which suggests those low doses are mostly safe, it may be a more accurate model means average adults are at say 1/2 the risk of linear models. But, say children are at 10x risk, babies are at 100x risk, and Fedus are at 1,000x higher risk than those linear models suggest. Thus because of a high risk to a significant population it’s even more dangerous.

It’s very easy to find some way a specific model is wrong in one direction, but hard to say if that’s the only adjustment you need to make. Ex: Counting better cancer treatments making nuclear safer while ignoring longer lifespans making it more dangerous.