Tim Sweeney calls Steam's AI disclosure rule abhorrent and a political statement

What Tim Sweeney said about Steam's AI disclosure

On 23 October 2026 Epic Games founder and CEO Tim Sweeney replied on X to a post attacking Steam's generative AI disclosure with one sentence: "Forcing developers to write a political statement to be used against them by the hate mob they're helping organize is abhorrent." By the time AI Lookout read it on 25 October, X counted more than 870,000 views, about 1,050 replies and 467 quote posts on that reply alone.

He was answering Matt Workman, who had written that "the premise of an AI Generated Content Disclosure is ABSURD in 2026" and called it embarrassing for Valve and Steam. GamesRadar+ notes that Workman's bio says he works in "Content & Vibes" at the generative media cloud company Fal.

Later that evening Sweeney widened the argument in two more replies to Jim Geduldick, who had questioned him. In the first, he wrote that age ratings tell parents what is in a game, while "these AI statements are just as partisan as mandatory political essays or forcing developers to list their voting history to drive cancel campaigns and public struggle sessions." In the second, he said there is "a real economic cost" to Valve's "forced political disclosures" and that, besides taking 30 percent, Valve is taking a potentially larger fraction of developers' revenue "by undermining their business with these AI hate campaigns."

What Steam's generative AI disclosure actually requires

The rule Sweeney is attacking lives in the third section of Steam's Content Survey, which every developer completes before a store page and build go to Valve's review. Valve says the section is not about efficiency gains from AI tools built into modern game engines. It covers AI used to create content that ships with the game and is consumed by players, which Valve lists as artwork, sound, narrative and localization, among others.

The survey splits that content in two. Pre-generated content is anything made with AI tools during development that ships in the game. Valve says it reviews that output the same way it reviews non-AI content, including a check that the game is not illegal or infringing and matches its marketing. Live-generated content is created by AI while the game runs. For that, developers must also describe the guardrails meant to stop the system generating illegal content, and Valve says it will not ship live-generated AI in Adult Only sexual content games at this time.

The resulting text appears on the store page. GamesRadar+ quotes the disclosure for Call of Duty: Modern Warfare 4 as "Our team uses generative AI tools to help develop some in game assets." Valve set out the approach in its AI Content on Steam announcement in January 2024, and the Steamworks page still includes a note that AI in games is rapidly evolving, that "many artists and creators are rightfully concerned about exploitation and misuse," and that Valve wants to hear from developers who think it is overlooking something.

Why the argument flared during Steam Next Fest

The thread started with a disclosure, not with Sweeney. On 22 October a games creator posting as JadeMist Gaming shared a screenshot of what they called one of the worst AI disclosures in "the sea of AI slop" at the current Steam Next Fest, in which an indie developer said they "obviously" had to use AI to write their game's code. "If you want to use AI, fine," the post said, objecting to the claim that indies have no choice. Workman quoted that post, and Sweeney replied to Workman.

GamesRadar+ reports that this Next Fest showcase carried more AI-aided games than any before it and nearly double the total demo volume of the June 2026 event, though it adds that low-quality AI games quickly sank once Steam's recommendation systems kicked in. It also points to people using AI to clone trending games almost overnight, and to developers cancelling demos to make copying harder. In that market the disclosure line is one of the few signals buyers can read before they pay.

How it fits Epic's own push into AI

Sweeney's position is consistent with what Epic has been doing. Epic recently released a vibe-coded Epic Games Launcher preview that it says it rebuilt from scratch with AI coding tools and that starts faster than Steam in tests it commissioned. Bloomberg has also reported that Epic is retooling development to be AI-first, with Sweeney describing a huge pivot underway.

Destructoid says Sweeney advertised the 2026 Epic AI MegaJam on 22 October. GamesRadar+ reports that the Unreal Engine game jam initially required AI use and promised unlimited agentic support before Epic walked that back to "you can use as little or as much AI as you want." Sweeney's objection to Steam's label is also not new. GamesRadar+ says he called Valve's approach nonsensical in November 2025, arguing AI will be involved in nearly all future production, and compared the disclosures to a "Scarlet Letter" in June 2026.

The case for keeping the label

Some of the pushback came as a single question. Geduldick asked why Sweeney was afraid of disclosure when Epic uses rating labels, as does the whole industry across games, film and television, and whether an Epic developer who chooses not to use generative AI would be treated less well. Sweeney's answer drew the line at age ratings: those exist for parents, while an AI statement, in his view, only serves people who want to boycott.

Destructoid's Pedro Peres takes the buyer's side. He writes that the disclosures warn customers about the technology so that, instead of wondering whether a voice sounds a little too robotic or the art too uncanny, they can know before spending money on something they might not support. GamesRadar+'s Austin Wood adds that much of the criticism of AI in games targets art, music and similar assets, while more developers accept it as a programming aid. That is roughly the line Valve's own survey already draws, since code assistants and engine tools are not the focus of the disclosure.

Not every veteran sees the technology Sweeney's way either. Nintendo's Shigeru Miyamoto recently compared generative AI to word processors and said copied work does not last, a more cautious framing than the one coming out of Epic this month.

What this means for developers and players

This is analysis, not a policy change. Nothing in Sweeney's posts changes the rules: Valve still requires the disclosure, and the Steamworks documentation read on 25 October shows the same three-part survey. What the exchange does show is that the two biggest PC game stores now take opposite public positions on whether buyers should be told about generative AI in shipped content.

For developers on Steam, the practical points are unchanged. AI-made art, audio, writing or translation that ships in the build has to be described. AI-generated content at runtime needs a description of the guardrails. Using an AI coding assistant is not what the section targets, though some developers choose to mention it anyway, as the Next Fest example shows. Sweeney's economic argument is that the label itself costs sales. Neither Sweeney nor the coverage offers sales data showing that effect, and Valve has not published any either.

For players, the label remains the only standard, store-level statement of how a game used generative AI. It is written by the developer, checked by Valve against the build, and visible before purchase. Whether it reads as information or as a political badge depends, as this week's thread shows, on who is reading.

What we don't know yet

Valve has not responded publicly to Sweeney's latest posts in any of the coverage AI Lookout read, and the Steamworks page shows no change. Sweeney did not say whether Epic would ever add a comparable label to its own store, or how a developer should tell buyers about AI use if not through one. The impression and reply counts in this article are X's own figures as read on 25 October and will keep moving; they measure attention, not agreement.