-O3 gained a reputation of being more likely to "break" code, but in reality it was almost always "breaking" code that was invalid to start with (invoked undefined behavior). The problem is C and C++ have so many UB edge cases that a large volume of existing code may invoke UB in certain situations. So -O2 thus had a reputation of being more reliable. If you're sure your code doesn't invoke undefined behavior, though, then -O3 should be fine on a modern compiler.
Oh, there are also plenty of bugs. And Clang still does not implement the aliasing model of C. For C, I would definitely recommend -O2 -fno-strict-aliasing
That's a little vague, I'd put that more pointedly: they don't understand how the C and C++ languages are defined, have a poor grasp of undefined behaviour in particular, and mistakenly believe their defective code to be correct.
Of course, even with a solid grasp of the language(s), it's still by no means easy to write correct C or C++ code, but if your plan it to go with this seems to work, you're setting yourself up for trouble.
Flags from -O3 often flow down into -O2 as they are proven generally beneficial.
That said, I don't think -O3 has the problems it once did.