Do bots have too many rating points?

So today, I played a game with the chess computer "Wally". The game ended with white (me) winning by checkmate in 57 moves. When I went to check the analysis, I found that I played with an 80-something percent accuracy. While this is good, regardless of the opponent, I was wondering how I beat an 1800-rated chess computer. Keep in mind that I'm only 660 rapid, 750 blitz, and 200 bullet (I suck at speed chess). This means the chess bot supposedly has more than twice my level of chess institution. This led me to think that these bots are "overrated", and should be matched with a lower rating. There are, however arguments against this, so maybe this will turn into a debate?

Here's the game:

yes, they are overrated

I have a game for the PS2 called Chessmaster. It has bots whose ratings change along with your in-game rating when you play them, and also when they play against each other in tournaments. They are over-rated, but a quirk in the game's design, which may also exist in the real Elo system, idk, is that 1 Elo point is deleted every time a game is played, whether it's between you and a bot or between two bots, so the elo ratings, which come to represent the actual relative strengths of the bots, also deflate over time. It's possible to reset all the AI ratings to their original, over-inflated score, but I think it's fun to let the ratings deflate and come to represent the actual playing levels of the different bots. It's an outstanding game! If you still have a PS2, you should look for it, because it has strong bots and a lot of great learning material, too.