So far, Valve, the game developer and owner of the Steam platform, has only revealed a few clues about its involvement in generative AI applications. But if the recent findings are true, the company may be developing an AI function called "SteamGPT" to combat cheating on the Steam platform and manage account bans.
Gabe Follower, a data miner who specializes in Valve-related content, has long been digging out new information by tearing apart Steam code updates. He recently discovered references to "SteamGPT" in the code, and the relevant content is directly related to account data statistics and Steam trust score indicators. This system is Valve's core tool for dealing with cheating in Steam online games and enforcing account ban restrictions.
According to Gabe Follower, this code seems to include multiple account data as variables, including account registration time, credibility score, model evaluation results, and even the account's past ban record. If Valve is indeed developing this AI-assisted tool, its goal is most likely to improve the efficiency of processing customer service tickets and solve recurring problems such as incorrect closures.
Valve has been troubled by cheating issues on the Steam platform before, and its technical methods are not always able to cope with it. At the end of 2023, Valve's anti-cheating system (VAC) experienced an incident of mistaken ban. Players were banned just for moving the mouse too fast in "Counter-Strike 2". At the same time, the problem of cheating has always been a stubborn problem that Valve needs to solve urgently, so AI automation tools are quite attractive to it, even if this controversial technology is always accompanied by many doubts.
Nonetheless, the code unearthed this time may just be content that Valve is testing but will not eventually be launched. Even if this feature is eventually launched successfully, "SteamGPT" will most likely be used as an internal tool rather than a feature open to ordinary consumers.