Should the data be collected without consent by exposing the public to it for decades or provided by clinical trials funded by the company that wants to bring it to market?
You are asking for the impossible. It can be impossible to tell with things that only have no - to mild effects. Proving a negative ("this is not dangerous") is hard :)
> It can be impossible to tell with things that only have no - to mild effects.
If any effects are unmeasurable with standard clinical trial practices, then it's safe. People care about things that actually are harmful.
> Proving a negative ("this is not dangerous") is hard :)
Which is why I never asked for it. Safety standards are always phrased positively. OSHA doesn't say "workplaces must not be dangerous", they say "wear a hard hat, the employer shall do ___, employees must not be forced to do ___". It's very easy to regulate that "new compounds must be shown not to affect human hormone levels at X concentration, must be shown not to be carcinogenic in Y model of cancer" etc.
Even still, PFAS started being used in the 50s. If it took us this long to figure out there is something dangerous about it, chances are a small study would not have found anything either. We’ve had lots of time and a huge N.
You collect data on a smaller scale, like clinical trials do for medications. There's a big range between "no data" and "we've enrolled eight billion humans in an uncontrolled trial".