Many, maybe most, of the wilder claims in science (by scientists) are made to bring attention and funding to a field. Science journalists are often happy to join in because surprising results sells more papers and ads.
As fundamental discoveries are made, future work tends to be more peripheral, even if important. For example, microbiomes clearly influence the host organism, but the discoveries in this field are not on par with understanding the structure/function of DNA. Germ-free mice are recognizably mice, DNA-free mice are not.
The problem is when other scientists do literature searches,
Then the next problem is that journalists find something interesting and publish it as fact, and so both laymen and scientists are lead astray.
The problem is when other scientists do literature searches, find the "answer" then base their entire work off of it. Every scientist after that does the same thing and suddenly a wild-ass-guess becomes fact.
A great example is in epidemeology. I work in biotech and many of our forecasts are based on epi data. I remember one time when someone questioned a numbered, so our epi team did some work. The number was the only estimate available, but it turned out to come from a paper from the 1700's that looked at the prevalence of this disease in orphan children in some small Frech town. We would have been better off just guessing.
This happens a lot.