Anti-cheat methods in mainstream games are becoming quite interesting with the use of machine learning / surprisingly advanced defensive AI. People trying to auto-aim or even create new accounts for boosting in Overwatch are getting banned in literally half a game (5 minutes)

Blizzard records every single game that's ever played - millions - including how every player moves, strafes, zooms, runs, which parts of the map they occupy. It can differentiate between an extremely nuanced set of skill brackets and instantly tell when something is not right.

We've moved well past the age of detecting duplicate accounts based on IP / similar names / random OSINT. Now we're into "we're gonna analyse every single tiny thing you ever do and match it against every single tiny thing anyone else has ever done". This will apply to IRL soon.

I use Overwatch as an example a lot of the time when it comes to AI - for example, Google can easily win at Chess/Go and even Starcraft. But if it wins a 6v6 match against a pro Overwatch team, that's absolute insanity.

The thing about this game is that even if the AI has 100% aim accuracy and map awareness, the on-the-fly human element is so complex that, without field of vision of the entire map, an AI would have to genuinely develop some sort of "natural curiosity" to counter every action.

I ran an experiment with Overwatch where I created a new account and played a single ranked match, deliberately moving and shooting like a new player would. I then started aiming better and using smarter map awareness and was permabanned within 60 seconds, mid-match. Goddamn AI.

Anyway those are my morning thoughts on game anti-cheat anti-smurf anti-booster AIs. Wait until someone like Google gets their hands on that dataset and does something ridiculous with it.