I don’t know if we have a thread or any other avenue of reporting these, but in this game autoscore seems to have messed up J4 stone:
Since it’s a bot game, there is no recourse. On the other hand, it didn’t change the result of the game. Is it the case that the autoscorer is more likely to not bother scoring correctly if it doesn’t change the result?
Yes, that is the Autoscore Bug. It has been known since August 2020. I’ve written about it a lot in various threads. One of the CM threads has the fullest discussion, but I don’t remember which one, and I’m tired of the whole subject.
It’s a little different from the normal cases in which the players’ browsers are expected to trigger autoscore! With bot games somehow the server handles it, eliminating any possibility that the problem lies with the players’ network connections.
Periodic reminder that bots, via the GTP protocol, can mark dead stones and come to agreement of the status of stones during scoring phase, just like humans. It is OGS’s decision to not use this feature that leads to the automatic application of autoscore on bot games with no chance for humans to correct autoscore mistakes, and the subsequent frustrations, forum posts, and moderator workload that causes.
I think if you make all bot games unranked then you can do anything you want with bot games and scoring.
If you want ranked bot games, it doesn’t make sense to have someone argue with a bot till it breaks or accepts the wrong score.
Or if you’re forced to play on until the bot accepts the score you want it to. Imagine a bot arguing that it’s stones aren’t dead like at https://playstrategy.org/ where the bots won’t accept any stones as dead ever.
It’s just a whole other set of problems and reports to deal with instead.
KGS has ranked bot games without all these problems OGS has (there’s no autoscore, you need to mark and agree on dead stones). A bot and a human not coming to an agreement on the status of the stones is as much of a problem as a human and a human not coming to agreement on the status of stones.
Treat them like humans who score cheat, a warning and instruction to bot operator to fix their bot, and ban if they don’t.
I legit thought this was a necro as i started reading your comment 
It looks like I need to clear up a misapprehension that has arisen in this thread about the nature of the Autoscore Bug. It doesn’t seem to have anything to do with bot games per se. Indeed, I don’t recall a single example from the past of it occurring in a bot game. (That could be a sample bias, because I never watched bot games, and bot games don’t get reported for cheating.) However, let’s be clear, the Autoscore Bug has been wrecking a significant number of human games for nearly five years now.
We’ve just moved the frustration and the workload somewhere else.
It’s great if KGS can handle it, maybe we can find out if it’s because KGS has no cheaters, or if its rating system is more stable.
The thing is noone is watching you cheat against a bot, the bot probably won’t be the one reporting you. If you cheat in 100s of games against the bot, only someone paying attention and noticing the bots rating tanking will end up reporting it.
Sometimes bot operators might be the ones paying attention and report cheaters, but again it’s possibly hundreds of games to have to annul regularly, with bots that could play hundreds of games a day.
At least the current types of cheating against bots will be fairly repetitive and boring, and if it’s an exploit, it’s something you’d have to figure out.
If it’s just a case that a bot lets the human decide the scoring, or gives up after a certain number of disagreements, then there’s a fairly easy way to beat the bot every time.
I’m not sure why you’d sign up for that, over the small trickle of people that have issues with the current autoscoring, which can still be improved over time.
I’d say BadukPop gets a similar amount of posts about how it’s autoscoring games, but I doubt they’ll swap to letting you argue scoring with their bots any time soon.
They don’t even have manual scoring of human games. Same with GoQuest, no manual scoring of human games even.
I also don’t think being able to argue with a bot is a good user experience feature, unless that’s meant to be enjoyable.