If you check for 10 patterns, each at the 95% significance level (p<0.05), there's a 40% chance of getting at least one false positive. Look for 20 patterns and that goes up to 60%.
That's why science requires you to formulate a hypothesis before running the experiment. And why finding a pattern through data analysis does not warrant a witch hunt.
95% significance of the fact that no women or ex convicts were hired for the job? Seems significant enough that a newspaper would write about it.
95% significance of not hiring a guy who looking too much of a hipster and has too many tattoos? Not sure, to be honest - I can see how society (and newspaper writers) would easily dismiss that as bs.
Don't try to abstract it away as otherwise it just seems you're justifying something absolutely ridiculous.
It also isn't about setting up a witch hunt. It is about protecting people from discriminatory practices even when the discrimination is neither overt or even intentional.
The existence of that pattern by itself might be enough for a company to be guilty of discrimination. If a policy disproportionately harms a specific protected class and they can't show a legitimate business reason for that policy, it is illegal regardless of intent. It is called disparate impact.