96.9 accuaracy

Is that normal for a human?

anyone can have a good game lol

Why not? I had this a couple of times.

Depends a lot on skill level. For a GM it is very possible but most of us are unlikely to be that accurate in any particular game, although anyone can have a good game and get that accuracy occasionally, especially if the game is straightforward and the moves are easy, like if the opponent makes blunders with simple responses. Is there a reason you ask about this specific number?

I’m 1300 and had a couple of games with no mistakes or blunders and maybe 1-2 inaccuracies. I’ve no idea what the % was as I never checked.. but as someone else said, perhaps your opponent played well and you didn’t. If someone who isn’t a top player is regularly achieving that level then something could be afoot. I’m obviously no expert but that’s my take based on what I’ve read in other posts and from videos I’ve watched. I will add that that’s 1-2 games from about 1200 games.

Literally a 1300 got this twice in a row during one of my tournaments

It depends. If I use my personal database together with the books I often reach this in cc games without engine. The reason lies in the opposition. There are lines easier to play or where I have theory till move twenty+ very often. Knowing the structure is a big advantage.

If the opposition is stronger, the accuracy level goes down often with the length of the game. Let's take a special line:

1.e4 e5 2.f4 d5 3. exd5 exf4 This is much easier to play for black. White(!) has only a narrow path to equalize. So black has often a high accuracy.

There are more examples. If you play the Khalifman books for white similar things happen often.

In rapid or classical I will be much lower, as you can see with my games. Analyzing and checking motives in the position helps a lot.

well let's say that if your opponent gives up a piece in the first 15 moves and the opening was played more or less perfectly, then the accuracy would even be higher than 97%. I had a game against a 2200+ in blitz where my opponent resigned after 12 moves due to the fact that he fall into a common opening trap and thus I had 16 of ACPL (which I find quite given the length of the game). Also, some strong players (2500+) would be able to dominate his/her opponent if they commit an early blunder (giving up material, a big positional advantage, damaging the pawn structure....) then the accuracy would be extremly low. I may also add that in official OTB tourneys, strong players rated 1800F+ would have games where they wouldn't do any single mystake as they have 90min per player and thus more time to think for ideas, plans, which move is the best .... And that's why I find slow chess amazing because some players manage to find magnificent combinations on the board.

Accuracy can be highest for the weakest players - as if someone blunders a queen after 5 moves, the algorithm will treat the many resulting moves that keep the +10 advantage as 'perfect', even if they result in long and convoluted approaches to winning the game.

It depends. If someone is having only 95%+ games then that's suspicious.
But you should consider mostly the quality of moves and time spend of them. If for example someone is taking a same time thinking about an obvious recapture or a genius move, then it's 99% cheater.
Maybe you just show us the game? Because it's hard to tell by accuracy only.