How Steam DESTROYS AI Slop Games… By Doing Nothing

Summary

The video argues that Steam limits the impact of AI-generated “slop” not by banning it, but by allowing games onto the store and relying on market incentives and player behavior to determine which ones succeed.

Steam’s AI policy and the range of games

• Valve initially rejected some games containing AI-generated art in 2023, citing uncertainty about rights to AI training data. In 2024, it shifted to requiring developers to disclose how they use AI, so players can make their own decisions.
• The scale has grown quickly: the video cites estimates that roughly one in three new Steam games now includes an AI disclosure. But an AI label does not necessarily mean a game is low-effort.
• Examples of different outcomes include:
  • South Scrimshaw, a largely handmade game whose narrator uses AI-generated voice, which has very positive reviews.
  • Zup’s Zero, which discloses using AI for a small amount of footage and also received strong player reviews.
  • Creonol and several expensive AI visual novels, presented as examples of low-effort games seeking quick profits.

Why strict gatekeeping struggles

Nintendo, Sony, and Microsoft review games before release, but the video says that approach has not stopped shovelware:

• Nintendo faced clones of Unpacking and changed its eShop charts to rank games by revenue rather than copies sold.
• Xbox’s store had many inexpensive achievement-farming games.
• Sony removed large batches of games from certain developers, while still facing waves of cheap releases.

The video compares today’s situation to the early-1980s video-game crash. A rush of new publishers and low-quality games helped flood the Atari market; Nintendo later restored confidence with tighter approval rules, a quality seal, and limits on releases.

The video argues that this model is harder to use today: Steam receives dozens of new games per day, and AI makes producing games faster and cheaper.

Steam’s approach and its five “filters”

Steam moved away from hand-picking games. After its Greenlight experiment, Valve introduced Steam Direct in 2017, allowing developers to publish games for a $100 fee, refundable if the game earns more than $1,000. Valve says it mainly excludes games that are illegal, malicious, or outright trolling.

The video describes five forces that can make low-effort games unprofitable:

1. The $100 publishing fee: It creates a cost for developers who want to release large numbers of games, and they only recover it if a game reaches the required sales threshold.
2. Steam’s recommendation system: It relies on what players actually do. Games that attract little play are less likely to be recommended, making visibility difficult for games that depend on mass exposure.
3. Player reviews: Players can quickly warn others about poor-quality games. The video cites research finding that games with AI disclosures tend to receive fewer reviews and are less likely to reach 100 reviews, while noting that AI use may overlap with low-budget, low-effort projects.
4. Refunds: Buyers can request refunds within two weeks if they have played for less than two hours, limiting the money a disappointing game earns.
5. AI disclosures: Steam labels AI use and leaves the judgment to players. The video says most surveyed players would not automatically reject an AI-assisted game if it were good.

Together, these mechanisms mean that AI-disclosed games make up a substantial share of releases but a smaller share of estimated sales. The video’s central point is that a game can be on Steam without being visible, played, or profitable.

Where “doing nothing” fails

• Block Blasters was initially a normal game, but a later update reportedly introduced malware that stole cryptocurrency from players. One victim was streamer Rasselant, who lost funds raised for cancer treatment after downloading the game during a livestream. Valve removed the game after the incident became public.
• The video also describes FBI investigations into malware distributed through games on Steam. Unlike ordinary low-quality games, malware can cause harm before reviews, refunds, or recommendations have time to warn players.
• Other limits include uncertainty about whether developers accurately disclose AI use, the lack of a built-in option to filter out AI games, and the risk that a huge volume of releases also buries worthwhile indie games.
• The video notes that Steam removed some adult games after payment processors warned that they might violate card-network or banking rules, showing that the platform cannot leave every decision to players.

Overall conclusion

The video’s verdict is that Steam’s open approach does a better job of making AI slop commercially ineffective than strict moderation does. Players’ choices—what they play, review, and refund—help determine which games get attention. But this works best when the main risk is wasting time; when a game can steal money or distribute malware, the video argues that Valve needs to intervene.

People, sources, and companies mentioned

• People: Nick (the video’s presenter); Jim Sterling; Howard Scott Warshaw; Tim Sweeney; ZachXBT; streamer Rasselant.
• Publications and research sources: PC Games; PC Gamer; TrueAchievements; Kotaku; Game Analytic; SteamDB; the FBI; and the cited Steam-player survey and game-release analyses.
• Companies and platforms discussed: Valve/Steam, Nintendo, Sony, Microsoft/Xbox, Epic Games, Deezer, and Spotify.