I don't know what chess.com's algorithm for deciding on "brilliant" moves is exactly, but your "simple king-side castle" is in fact close to "some insane sacrifice" that gives a small advantage.
You see, your 4.Ng5 was a blunder that gives a piece away for nothing, black should just take it with 4...Qxg5.
However, after 4...Nh6 5.0-0, the piece is still en prise, but now 5...Qxg5 doesn't work because of 6.d4! and now white wins the piece back, and has a small advantage because of the bishop pair and the slightly odd positions of the black queen and knight on h6.
After 4...Qxg5, 5.d4 didn't work because black had 5...Qxg2, attacking the rook, so he can move the bishop to safety after white plays 6.Rf1.
So your simple king side castle makes your previous blunder actually work, because it covers pawn g2.
[FEN ""]
1.e4 e5 2.Nf3 Nc6 3.Bc4 Bc5 4.Ng5?? Nh6?? (4...Qxg5 5.d4? Qxg2 6.Rf1 Nxd4) 5.O-O! Qxg5 6.d4! Qh4 7.dxc5
Qxg5
withd4
. The platform didn't know you simply blundered.Qxg2
)