Releasing the exploit, on the other hand, is a different story (don't do that)
Per my understanding, there is a distinct difference between releasing a bug (that you found in software on your own machine) vs. releasing a tool that exploits that bug.
This also comes up in discussions about exploit markets: you can sell an exploit to a bounty program, or to Zerodium or whatever, but people always think you can make more (for instance, if it's some dumb SQLI, more than the $0 the public exploit markets will pay you for it) by selling "on the dark web". But if you're selling a bug that only lets you exploit some specific SAAS application "on the dark web" to some anonymous criminal (or someone a jury will think you should have known was a criminal), you're taking a chance that a prosecutor will make a case that you took money to facilitate a specific crime that they're now going to charge you with.
My example was that cracking a game is illegal, but releasing a cracker is much much worse.
And, again, IANAL.
I think publicized security research --- which I obviously support --- has created expectations among technologists that either weren't intended, or were more wishful than the facts support (lots of researchers, most probably, think the CFAA is a bad law and are happy to set expectations premised on its illegitimacy, but: see Wesley Snipes for how that turns out).
If there weren't lots of public security research, I don't think you'd really have to tell developers not to go looking for vulnerabilities.
Again, and to be clear: if it's something running on your own machine, or a machine you own, go nuts.