Generative AI’s Growing Footprint in Steam Games

Steam’s disclosure system for AI-assisted content has quietly created one of the best public lenses into how game studios are actually using generative AI. Recent analysis of those disclosures suggests that games openly admitting to AI usage have already generated an estimated $660 million in gross revenue on Valve’s platform, spanning everything from small indie experiments to headline releases like Call of Duty: Black Ops 6 and beloved strategy titles such as Stellaris.

This does not mean $660 million worth of games are fully AI-generated. Instead, it signals something more subtle and more important: generative tools are becoming routine parts of the development pipeline, even in large productions, and players are only just starting to see what that looks like.

Why Steam’s AI Disclosures Matter

Valve introduced AI disclosure requirements to address questions around copyright, safety, and transparency for games that use generative systems during development or at runtime. Developers must now answer whether generative AI was used and, if so, in what way. In practice, this has created a self-reported database of titles whose creators are willing to acknowledge AI use.

A breakdown of these disclosures shows:

• Thousands of Steam games now carry some form of AI disclosure.
• Some of those games belong to well-known franchises with large marketing budgets.
• The vast majority of AI use is behind the scenes, supporting artists, writers, and designers rather than replacing them outright.

While the disclosure system is imperfect—developers can make mistakes or choose conservative wording—it still offers rare visibility into how widespread generative tools have become across the PC gaming market.

From 1,000 to Over 10,000 AI-Disclosing Titles

The scale of AI use on Steam has grown rapidly in just a short time. A former Steam Labs developer, Ichiro Lambe, estimated in 2024 that around 1,000 games on the platform explicitly mentioned using generative AI in some capacity, ranging from small indie projects to mid-budget multiplayer shooters.¹ As Valve formalized its AI policies and disclosure prompts, that number multiplied.

Subsequent independent analysis found that by 2025–2026, more than 10,000 titles on Steam featured some form of AI disclosure, representing roughly 8% of the entire Steam catalog at the time and over 20% of new releases for 2025 alone.² These figures are not official Valve data, but they align with broader industry surveys from the Game Developers Conference (GDC), where a growing share of developers report actively experimenting with generative tools.³

These numbers reflect any type of disclosed AI use, not just flashy in-game features: a visual novel that used a model to polish dialogue and a strategy game that uses AI for dynamic storytelling both count.

The $660 Million Question: How Revenue Was Estimated

The widely cited $660 million figure for Steam games with AI disclosures comes from combining disclosure data with a simple revenue estimation approach sometimes called the “Boxleiter method.” This method, used by some analysts and developers, approximates a game’s gross revenue on Steam by multiplying three public factors:

• User review count (as a proxy for unit sales)
• Base price (or typical selling price)
• A constant factor (often 35) to account for the ratio of reviews to purchases

In formula form:

Estimated gross = Review count × Price × 35

When this rough model is applied to all games that voluntarily disclose generative AI usage on Steam, the combined gross lands around $660 million. The estimate is deliberately conservative and acknowledges several uncertainties:

• Some players never leave reviews, inflating the constant.
• Sale pricing and regional pricing complicate any single multiplier.
• Free-to-play titles with in-game purchases are poorly captured by this approach.

Nonetheless, the figure is directionally useful. Even if the actual revenue is significantly higher or lower, the conclusion is the same: AI-using games are not fringe curiosities. They collectively represent hundreds of millions of dollars in customer spending.

How Revenue Is Distributed Across AI-Using Games

One of the more telling findings from the analysis is how uneven the revenue distribution is among AI-disclosing titles. The vast majority of these games make very modest sums, while a small group of hits brings in most of the money.

Using the estimated revenue brackets, the landscape looks roughly like this:

• 12 games are estimated to have grossed at least $10 million.
• 33 games land somewhere in the $1–10 million band.
• 170 games reach six-figure gross revenue.
• 487 games sit in the five-figure range.
• 9,500+ games earned less than $10,000 each.

In other words, most AI-disclosing games are small projects or niche titles, but the mere existence of a dozen eight-figure earners signals that generative tools have already arrived in high-budget, highly visible productions as well.

What Big Names Like Call of Duty and Stellaris Are Doing With AI

The inclusion of blockbuster franchises among AI-disclosing games has attracted the most public attention. When a series like Call of Duty or a long-running strategy title like Stellaris references generative AI, players naturally wonder whether their favorite characters, worlds, or voice lines are being shaped by algorithms.

In practice, disclosures from large studios often point to specific, constrained uses of generative tools rather than fully AI-authored content. Common patterns include:

• Support tooling for writers – Using language models to brainstorm variations of dialogue, quest text, or lore snippets that are then edited by human narrative designers.
• Concept art and ideation – Relying on image models at the earliest art stages to explore mood, composition, or costume variants, followed by fully manual production art.
• Localization aids – Employing AI-assisted translation engines as a starting point, with human translators and editors finalizing the text for each market.
• In-game dynamic content – Experiments with enemy barks, flavor text, or background chatter generated or recombined on the fly, usually within strict guardrails.

Major publishers are acutely aware of the reputational and legal risks associated with AI-generated assets, especially in light of ongoing court cases and complaints around training data and copyright.⁴ As a result, they tend to keep disclosed AI use tightly scoped and emphasize that core creative decisions remain in human hands.

What Generative AI Is Actually Doing in Most Steam Games

While high-profile names draw attention, the majority of AI-disclosing games on Steam are small or mid-sized projects using generative tools in even more modest ways. Developers frequently describe AI as another entry in the long list of production tools, similar to procedural generation, physics engines, or auto-rigging software.

Common use-cases include:

1. Art & Asset Support

• Creating early placeholder art during prototyping.
• Generating variations of icons, textures, or background elements.
• Producing reference material for human artists to reinterpret.

2. Writing & Editing

• Polishing or rephrasing dialogue to improve clarity and tone.
• Generating descriptions or item flavor text in bulk, later edited by writers.
• Brainstorming quests, branching options, or character backstories.

3. Localization & Accessibility

• Draft translations for multiple languages, then refined by native speakers.
• Assisting with subtitle generation or closed captions.
• Creating prototypes of text-to-speech voiceovers before full recording.

4. Gameplay & Systems

• Procedural quest generation that mixes AI text with hand-authored templates.
• NPC dialogue systems that recombine human-written lines using AI rules.
• Dynamic hint systems that respond to player behavior more flexibly.

Across these examples, the pattern is the same: generative AI rarely replaces an entire role. Instead, it compresses certain steps, allowing small teams to aim for broader scope or more content than they could otherwise support.

Valve’s Evolving Rules: What Must Be Disclosed

Valve’s stance on generative AI has shifted from cautious gatekeeping to conditional acceptance. In early 2024, the company caused debate when it paused and then later allowed games that used generative tools, setting expectations around legality and content safety. By 2025, its policies focused on two main principles:

1. No illegal or infringing content. Developers must ensure that any AI-generated material does not violate copyright or other laws.
2. Clear disclosure for certain AI uses. Games must flag AI-generated or AI-assisted assets in their store pages when those assets appear in the shipped product or in marketing materials.¹

Recent reporting and community discussions indicate that Valve has relaxed disclosure requirements for purely internal uses. If a team only employs AI for tasks like coding assistance, internal documentation, or non-shipped prototypes, they may not need to check the disclosure box at all. However, if AI directly contributes to shipped art, audio, or text, or if it’s used for promotional screenshots, trailers, or descriptions, disclosure remains required.

This distinction means the actual number of Steam games touched by AI is almost certainly higher than the disclosed figure. Many studios now use AI-powered coding tools or writing assistants internally, without any direct AI-generated content in the final game.

Industry Context: Developers Are Rapidly Adopting Generative Tools

Steam’s disclosure data aligns with a broader trend captured by industry surveys. The GDC State of the Game Industry report for 2024 found that around 38% of surveyed developers were experimenting with generative AI tools for prototyping or iteration, and an additional group expected to explore them soon.³ Major engine vendors and platform holders have also started integrating AI features into their ecosystems.

For example:

• Large publishers have announced in-house AI platforms for aiding scriptwriting, animation, and QA testing.
• Mobile analytics firms such as Sensor Tower now track generative AI app trends alongside game revenue and user acquisition patterns, highlighting how AI-focused apps and AI-assisted games compete for attention and spending.⁵
• Indie studios increasingly mention AI tools in postmortems as a way to cope with limited budgets and small teams.

Within that context, $660 million in estimated Steam revenue tied to AI-disclosing titles looks less like an anomaly and more like an early snapshot of a change that is still accelerating.

What This Means for Players

For players, the rise of AI-disclosing games raises several intertwined questions about quality, ethics, and transparency.

1. Quality and Authenticity

Some players worry that generative content may feel generic, repetitive, or less emotionally resonant than human-authored work. Others argue that when used thoughtfully—especially for systemic storytelling or reactive NPCs—AI tools could enable kinds of gameplay that would be impractical with traditional pipelines.

2. Ethical and Legal Concerns

Generative models are frequently trained on vast datasets scraped from the web, including art, text, and audio created by people who never consented to that usage. Ongoing lawsuits have challenged these practices, forcing both AI vendors and studios to think more carefully about training data and licensing.⁴ Some players now actively avoid games that use certain AI tools, while others emphasize the need for clearer labels and more robust opt-out mechanisms for artists.

3. Transparency and Choice

Steam’s disclosure tags give players a basic signal but rarely specify how AI is used. Over time, we may see store pages adopt more granular labels—for instance, distinguishing between AI-assisted development and AI-generated in-game content. This would help players make more informed decisions based on their own comfort levels.

What This Means for Studios

For developers and publishers, the central question is not whether AI will enter their workflows, but how and under what constraints they choose to deploy it.

• Smaller teams can use AI to handle repetitive tasks, freeing humans to focus on unique, high-impact work.
• Larger studios can experiment with dynamic content and more personalized experiences, but they must manage reputational risk and negotiate with unions, including voice actors and writers concerned about automation.
• Publishers must balance cost savings against long-term brand trust, especially when players react strongly to perceived “corner-cutting” via AI-generated assets.

The early revenue numbers suggest that openly disclosing AI use has not doomed these projects commercially. Many AI-disclosing titles have found strong audiences. However, the debate around fair training data, job security, and artistic integrity is far from settled.

Looking Ahead: Will AI-Disclosing Games Become the Norm?

Given current growth trends, several analysts now predict that by the end of 2026, one-third of new Steam releases will carry AI disclosures of some kind.² Combined with the fact that internal, non-disclosed AI use is already widespread, it is reasonable to expect that almost all mid-sized and large studios will be using generative tools in at least some capacity within a few years.

Whether players view that shift as a threat or an opportunity may depend on how responsibly studios handle the transition. Clearer disclosures, stronger licensing frameworks for training data, and more visible commitments to preserving human creativity will likely shape how accepting audiences are of AI-powered game content.

For now, the $660 million milestone is less a destination than a signpost. It shows that generative AI is no longer a futuristic promise; it is already woven into the business realities and creative decisions of the games people are playing today.

Frequently Asked Questions

Do AI-disclosing games on Steam mean the entire game is AI-generated?

No. In most cases, the disclosure simply means that some part of the development process used generative tools, such as placeholder art, localization drafts, or dialogue polish. The core design, narrative, and art direction are typically still led by human developers.

Can I see exactly how a Steam game used AI?

Not usually. Steam’s current tags indicate that AI was used, but they do not require a detailed breakdown. Some studios voluntarily explain their workflows in devlogs or community posts, but that level of transparency is not yet standard.

Why are big franchises like Call of Duty disclosing AI use?

Large studios face both legal scrutiny and community pressure. By disclosing AI use, they comply with platform requirements and try to control the narrative around how these tools are used—typically emphasizing that AI is assisting production rather than replacing human creativity.

Could the $660 million revenue estimate be wrong?

It is an approximation based on public data and a simple model, so it is almost certainly not exact. However, even if the real figure is significantly higher or lower, it still indicates that AI-using games collectively account for hundreds of millions of dollars in revenue on Steam.

Will all games eventually use generative AI?

Not necessarily, but many developers expect AI-assisted tools to become as common as physics engines or procedural generation. Some creators will continue to avoid generative systems on ethical or artistic grounds, while others will adopt them selectively for specific tasks.