It seems with each passing day, cheaters develop new ways to seem legit. This time, cheaters have figured out how to use generative AI to impersonate real players in CS2. Not even Valve’s flagship shooter is safe from man-made horrors.
The exploit appears to have been in the game for some time, though it gained traction with an April 24 Reddit thread that shows an AI talking to other players in chat. The AI, most likely a large language model (LLM) similar to ChatGPT, even spoke in slang to appear more genuine. However, the players quickly caught on and asked it to create a Python script, which it did. From the context of the video, it’s difficult to determine if the player was real and made the AI respond instead of them, or if the cheater was a bot account.
Players responded as you might imagine. Some were horrified, others were impressed, and others were worried video games are on the brink of being invaded by vast armies of bots and LLMs. “This game is fucked ngl, there better be a good fucking plan,” one disgruntled player wrote. “Valve, wake the fuck up, please,” said another. There have recently been rumors of cheaters employing AI to create cheats (thanks BO3) that are nigh undetectable, since they wouldn’t read into your computer’s memory or run as a traditional software cheat. If those cheats evolve significantly, combined with AI language bots like in the video above, we might be facing a terrible future.
Bots have been an issue for Valve for a long time, starting with Team Fortress 2 being rendered unplayable for a while. CS2 even has its own issues with bot armies invading Casual lobbies. Valve hasn’t yet reacted to the general cheating problem plaguing CS2, but hopefully they’ll have a solution in store for the AI bots as well.
