To give you an intuition though, LNT would suggest that the body has no repair mechanism for healing damage from radiation (should be suspicious about this) and that all effects are accumulative (which makes sense if there is no repair, but doesn't if there is repair). We can heal from other things and have demonstrated in the lab that cells can repair from low radiation dosages.
Basically the thing is that measuring is incredibly difficult because there's a temporal component and effects are very different depending on where that dosage is received (eg. your eyes vs your hands. See equivalent dosage). So what do you do? You overestimate on the side of safety. Failure of modeling is built into the safety standards (I used to work on radiation shielding devices and I fully support the use of the LNT model in practice. Better safe than sorry).
I haven't read all of these and didn't archive what I have read so these are what I came up with quickly (I did read abstracts, of course)
(Short end is that it isn't clear how much cancer risk increases for dosages <100mSv -i.e. "low dosages" -, which is WELL above occupational standards - 5x actually)
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2663584/
https://www.sciencedirect.com/science/article/pii/S000927971...
not an article but a good high level response (notice the hand waviness) https://hps.org/publicinformation/ate/q12755.html
This part of the wiki page is also good and shows you the contrasting sides https://en.wikipedia.org/wiki/Linear_no-threshold_model#Cont... It also has links to other journal articles.
https://atomicinsights.com/science-falsified-no-safe-dose-hy...
Also
Do people also assume that EM radiation is a linear no threshold? Because we know it isn't true for ionizing radiation, but it is a great model to use in practice because it overestimates harm (which we'd rather over estimate than under).
I've read a few of those papers on the EM radiation on mice and they seem to assume a LNT model, which doesn't seem all that honest to me. A few of the papers I read didn't have great p-values either and had drastically differing rates of cancer development for radiation levels and sex (IIRC one big one had high cancer rates for low power, nothing for medium, and moderate cancer rates for high power. Which there was no explanation to this. But that might have just been one bad paper).