TL;DR we found these chemicals present but with only one cherry picked exception we make no quantifiable claims of potential harm. That exception is lead, and we used the strictest standard we are aware of to describe the "dose", also note that we assume that the product is ingested to come to our conclusion. We also use weasel wording to avoid explicitly claiming that there are any regulatory or legal violations.
As someone who has worked in compliance testing for tightly controlled software platforms, things like this piss me off. These problems have known solutions.
> Every use case we have is essentially “Here’s a block of text, extract something from it.” As a rule, if you ask GPT to give you the names of companies mentioned in a block of text, it will not give you a random company (unless there are no companies in the text – there’s that null hypothesis problem!).
Make it two steps, first:
> Does this block of text mention a company?
If no, good you've got your null result.
If yes:
> Please list the names of companies in this block of text.
I'm more worried about possible impacts of an increase in atmospheric CO2 on cognitive function. It probably isn't an easily measurable amount on individuals, but across the worlds population it may have a notable effect.