Public Steam disclosure research25,997 disclosures tracked
Where do games disclose AI use?
Search developers’ Steam disclosures, compare classified AI use across
games, and follow recorded changes.
This index adds cross-game classifications and statistics to
statements published on individual store pages.
Groups reflect automatic analysis of public disclosures; mature records are excluded. These counts do not measure AI use across all of Steam.
Missing or unclear statements cannot establish whether a game uses AI.
This game incorporates generative AI to enhance visual storytelling and character depth in the following ways:
Video Content: A portion of the ending cinematic utilizes footage generated by Sora 2, which can be identified by its watermark.
Character Illustrations: Three soldier profile photos displayed within the Personnel Search Center were created using ChatGPT (DALL-E).
We have used these tools to supplement the creative vision of our team and ensure a more immersive atmosphere for our players.
This game uses AI assistance in the following areas:
- Localization: Some translations were assisted by AI tools.
- Some promotional or UI images were partially generated or refined using AI tools.
All gameplay content, art direction, and core assets are created by human developers.
1. In-game artwork
Some visual assets (e.g., character/decor sprites, UI icons) are generated with Google Gemini and OpenAI ChatGPT
(image generation) and further edited, retouched, and composed by the developer in the final art pipeline.
2. Store page and marketing graphical assets
Capsule art, library hero, library capsule, and promotional images are generated with Google Gemini and OpenAI
ChatGPT, and further edited, retouched, and composed by the developer before upload.
3. In-game localization
Text translations are AI-assisted using Anthropic Claude, with developer review for tone and accuracy.
Runtime behavior:
- No generative AI models are bundled with or called from the shipped game binary.
- All AI-generated content is pre-rendered and baked into final assets at build time.
- No user inputs are sent to any AI service during gameplay.
This game's development used generative AI in three specific areas:
1. Source code: Portions of the source code were authored with assistance from large language model (LLM) coding tools.
2. Localization: In-game text translations for 10 languages (English, Simplified Chinese, Traditional Chinese, Korean, Russian, German, French, Spanish, Brazilian Portuguese, and Italian) were generated
using LLM-based tools from the original Japanese source text.
3. Visual assets (UI graphics and logo): Generative AI was used in some cases to assist with creating certain in-game UI icons and the game logo. These AI-assisted assets are limited to generic UI elements
(icons, indicators) and abstract logo designs, and do not depict real-world characters, brands, or copyrighted materials.
At combo 32, a full synth pad blooms into the void. That peak harmonic layer is generated by Google's Lyria 3 Pro music model, then mastered, EQ'd, and timed by hand to fit each track's BPM and key. The other audio layers — kick, bass, arpeggio — are synthesized live in the Web Audio API. All beat charts, timing windows, and gameplay are hand-designed. No procedural or AI-generated content drives play.
Some visual assets and audio elements in the game were created using AI-assisted tools and then refined and integrated into the final experience by the developer.
Loopfall uses generative AI for select shipped content:
• Music — created in collaboration with ElevenLabs.
• Sound effects — generated with ElevenLabs Sound Effects.
• Localization — store text and in-game UI strings are translated from English to 14 other languages via the Anthropic Claude API.
All AI-generated content is pre-rendered and baked into the build — no live AI runs at runtime. Every asset is licensed under each provider's commercial-use terms and was reviewed by the developer before release.
Nothing is outside the scope of AI usage. Images were generated with an AI workflow. Sounds are free and edited. Code includes consultation with free tier LLMs.
This game uses pre-generated AI content for its Steam store capsules and some in-game audio elements. These tools were used to support the creative process and maintain the game's aesthetic consistency.
Some visual assets (card art and character portraits) generated using Google Gemini image models. Some SFX partially generated via ElevenLabs where Unity Assets were not available. All licensed for commercial use. Game design, code, and text content are original.
This game uses generative AI tools in part for creating UI character images and music.
These assets are further edited and refined by the developer before being implemented into the final game content.
No AI-generated content is created in real time during gameplay.
his game uses generative AI in the following areas:
🎵 Music: Portions of the game's original soundtrack were composed using generative AI tools.
🎙️ Voice Acting: Character dialogue is performed using AI-generated voices. All dialogue, script, and performance direction were written and directed by the development team.
All AI-generated content was reviewed, edited, and approved by the developers before inclusion in the final game.
All visual assets were generated in part using Nano Banana. Some code was generated using Gemini. This project was done by a Solo Dev and few eternal assets were used except where noted.
Generative AI was used solely as a supplementary tool for music composition and a small number of non-core textures (such as background patterns and decorative assets). All AI-assisted content has undergone significant manual refinement and human oversight to ensure it meets our artistic standards and visual style. AI was used only to enhance development efficiency; all core character designs, key art, and narrative elements remain 100% human-authored.
Some graphical assets, musical assets, localization, and portions of code were partially created with the assistance of generative AI programs and large language models (LLMs).
After I personally finalized the design phase, some art assets were first rendered using generative AI tools, and I then manually refined and detailed them.
Similarly, certain background music pieces were developed with the help of generative AI tools.
Should you have any concerns about these methods, please feel free to experience the game according to your own preferences. Thank you for your understanding and consideration.
Our game uses pre-generated AI content exclusively for audio production. Specifically, we utilize generative AI tools to assist in the creation of our background music tracks and to generate voice-over (VO) dialogue for characters. All AI-assisted audio is pre-rendered during development and baked into the final game files. We do not use live-generated AI at runtime, and we do not use generative AI for visual artwork, 3D models, code, or the written narrative.
Flux Tower Defense uses generative AI in the following development areas:
• Store page artwork: Main, header, library, small, and vertical capsule images, library hero, library header, and page background were created using AI image generation tools with human direction, prompt iteration, and post-processing. The "FLUX TOWER DEFENSE" wordmark typography was added manually.
• Achievement icons: The 20 in-game achievement icons were generated with AI image tools using hand-authored prompts and manually curated selections.
• Localization: Non-English store page descriptions were translated with AI assistance and reviewed before publication.
The game's runtime contains NO generative AI. All gameplay systems (towers, enemies, traps, waves, pathfinding, damage, upgrades), UI, and in-game visuals are deterministic code rendered live via HTML5 Canvas 2D. There are no pre-rendered sprites, no AI-generated game content, and no AI inference during play.
The model creation and icon production in the game used a combination of AI and manual editing. All translations were generated by AI. The store page promotional images were generated by AI and then manually edited.
This title uses generative AI for part of its localization process.
AI-generated translation drafts are reviewed and edited by the developer before implementation.
No content is dynamically generated by generative AI during gameplay.
Developer Logo was created partly using AI, some basic environmental obstacle and enemy placements are randomly generated with an algorithm. There are RNG elements.
Some of Bitroca's visual assets, UI elements, and audio were created with the help of generative AI tools, then
reviewed and finalized by a human developer before shipping. All content is pre-rendered and baked into the build no AI content is generated during gameplay.
This game was made available in many languages thanks to generative AI. Creation of some assets was performed or aided by generative AI tools.
If you ever beat an open-world game, you probably used more power than was used to generate all the assets in this game, including the ones that went unused. If you 100%'d that game, you used much, much more.
AI-assisted tools were used in the creation of some background images included in the game.
AI-assisted translation was also used for the English and Korean localization. The translated text was reviewed and adjusted during development.
There is no content generated by AI in real time during gameplay.
Some visual assets in Triad were created with the assistance of AI tools and further refined by the developer.
Triad is a solo-developed project, and these tools helped make its creation possible.
— Raph, developer of Triad
This game utilizes AI-assisted technologies in its development process. Generative AI was used to assist in writing code, refining localized text, and generating certain background ambient music and sound effects. Additionally, AI tools were used to help conceptualize and refine the unique "Shadow Puppetry" and "Ink-wash" artistic style assets to ensure a high-quality visual experience.
The game uses AI voice generation tools to create audio voiceovers for partial narrative texts and dialogue. All underlying text and story are originally created by the developer.
Some game artwork was generated with the assistance of AI, including monster sprites and a portion of achievement icons. These assets were then manually reviewed and edited to fit the game's pixel-art style. Store page descriptions were co-written with AI assistance.
Generative AI is used to assist with localization/translation. All localized text is reviewed and edited by the developer.
Generative AI is used to generate steam capsule image and asset. Act as placeholder
- Visual art: A portion of the pixel-art tiles and select character sprites were generated using PixelLab AI (pixellab.ai), then composited, edited, and integrated by the developer.
- Audio: Background music tracks were generated with AI music tools, then exported and sequenced into the game.
- Code & narrative writing: Game code and the in-game chronicler text (Brom's journal entries) were drafted with AI assistance (Anthropic's Claude) and edited by the developer.
This game incorporates generative AI to enhance visual storytelling and character depth in the following ways:
Video Content: A portion of the ending cinematic utilizes footage generated by Sora 2, which can be identified by its watermark.
Character Illustrations: Three soldier profile photos displayed within the Personnel Search Center were created using ChatGPT (DALL-E).
We have used these tools to supplement the creative vision of our team and ensure a more immersive atmosphere for our players.
This game uses AI assistance in the following areas:
- Localization: Some translations were assisted by AI tools.
- Some promotional or UI images were partially generated or refined using AI tools.
All gameplay content, art direction, and core assets are created by human developers.
1. In-game artwork
Some visual assets (e.g., character/decor sprites, UI icons) are generated with Google Gemini and OpenAI ChatGPT
(image generation) and further edited, retouched, and composed by the developer in the final art pipeline.
2. Store page and marketing graphical assets
Capsule art, library hero, library capsule, and promotional images are generated with Google Gemini and OpenAI
ChatGPT, and further edited, retouched, and composed by the developer before upload.
3. In-game localization
Text translations are AI-assisted using Anthropic Claude, with developer review for tone and accuracy.
Runtime behavior:
- No generative AI models are bundled with or called from the shipped game binary.
- All AI-generated content is pre-rendered and baked into final assets at build time.
- No user inputs are sent to any AI service during gameplay.
This game's development used generative AI in three specific areas:
1. Source code: Portions of the source code were authored with assistance from large language model (LLM) coding tools.
2. Localization: In-game text translations for 10 languages (English, Simplified Chinese, Traditional Chinese, Korean, Russian, German, French, Spanish, Brazilian Portuguese, and Italian) were generated
using LLM-based tools from the original Japanese source text.
3. Visual assets (UI graphics and logo): Generative AI was used in some cases to assist with creating certain in-game UI icons and the game logo. These AI-assisted assets are limited to generic UI elements
(icons, indicators) and abstract logo designs, and do not depict real-world characters, brands, or copyrighted materials.
At combo 32, a full synth pad blooms into the void. That peak harmonic layer is generated by Google's Lyria 3 Pro music model, then mastered, EQ'd, and timed by hand to fit each track's BPM and key. The other audio layers — kick, bass, arpeggio — are synthesized live in the Web Audio API. All beat charts, timing windows, and gameplay are hand-designed. No procedural or AI-generated content drives play.
Some visual assets and audio elements in the game were created using AI-assisted tools and then refined and integrated into the final experience by the developer.
Loopfall uses generative AI for select shipped content:
• Music — created in collaboration with ElevenLabs.
• Sound effects — generated with ElevenLabs Sound Effects.
• Localization — store text and in-game UI strings are translated from English to 14 other languages via the Anthropic Claude API.
All AI-generated content is pre-rendered and baked into the build — no live AI runs at runtime. Every asset is licensed under each provider's commercial-use terms and was reviewed by the developer before release.
Nothing is outside the scope of AI usage. Images were generated with an AI workflow. Sounds are free and edited. Code includes consultation with free tier LLMs.
This game uses pre-generated AI content for its Steam store capsules and some in-game audio elements. These tools were used to support the creative process and maintain the game's aesthetic consistency.
Some visual assets (card art and character portraits) generated using Google Gemini image models. Some SFX partially generated via ElevenLabs where Unity Assets were not available. All licensed for commercial use. Game design, code, and text content are original.
This game uses generative AI tools in part for creating UI character images and music.
These assets are further edited and refined by the developer before being implemented into the final game content.
No AI-generated content is created in real time during gameplay.
his game uses generative AI in the following areas:
🎵 Music: Portions of the game's original soundtrack were composed using generative AI tools.
🎙️ Voice Acting: Character dialogue is performed using AI-generated voices. All dialogue, script, and performance direction were written and directed by the development team.
All AI-generated content was reviewed, edited, and approved by the developers before inclusion in the final game.
All visual assets were generated in part using Nano Banana. Some code was generated using Gemini. This project was done by a Solo Dev and few eternal assets were used except where noted.
Generative AI was used solely as a supplementary tool for music composition and a small number of non-core textures (such as background patterns and decorative assets). All AI-assisted content has undergone significant manual refinement and human oversight to ensure it meets our artistic standards and visual style. AI was used only to enhance development efficiency; all core character designs, key art, and narrative elements remain 100% human-authored.
Some graphical assets, musical assets, localization, and portions of code were partially created with the assistance of generative AI programs and large language models (LLMs).
After I personally finalized the design phase, some art assets were first rendered using generative AI tools, and I then manually refined and detailed them.
Similarly, certain background music pieces were developed with the help of generative AI tools.
Should you have any concerns about these methods, please feel free to experience the game according to your own preferences. Thank you for your understanding and consideration.
Our game uses pre-generated AI content exclusively for audio production. Specifically, we utilize generative AI tools to assist in the creation of our background music tracks and to generate voice-over (VO) dialogue for characters. All AI-assisted audio is pre-rendered during development and baked into the final game files. We do not use live-generated AI at runtime, and we do not use generative AI for visual artwork, 3D models, code, or the written narrative.
Flux Tower Defense uses generative AI in the following development areas:
• Store page artwork: Main, header, library, small, and vertical capsule images, library hero, library header, and page background were created using AI image generation tools with human direction, prompt iteration, and post-processing. The "FLUX TOWER DEFENSE" wordmark typography was added manually.
• Achievement icons: The 20 in-game achievement icons were generated with AI image tools using hand-authored prompts and manually curated selections.
• Localization: Non-English store page descriptions were translated with AI assistance and reviewed before publication.
The game's runtime contains NO generative AI. All gameplay systems (towers, enemies, traps, waves, pathfinding, damage, upgrades), UI, and in-game visuals are deterministic code rendered live via HTML5 Canvas 2D. There are no pre-rendered sprites, no AI-generated game content, and no AI inference during play.
The model creation and icon production in the game used a combination of AI and manual editing. All translations were generated by AI. The store page promotional images were generated by AI and then manually edited.
This title uses generative AI for part of its localization process.
AI-generated translation drafts are reviewed and edited by the developer before implementation.
No content is dynamically generated by generative AI during gameplay.
Developer Logo was created partly using AI, some basic environmental obstacle and enemy placements are randomly generated with an algorithm. There are RNG elements.
Some of Bitroca's visual assets, UI elements, and audio were created with the help of generative AI tools, then
reviewed and finalized by a human developer before shipping. All content is pre-rendered and baked into the build no AI content is generated during gameplay.
This game was made available in many languages thanks to generative AI. Creation of some assets was performed or aided by generative AI tools.
If you ever beat an open-world game, you probably used more power than was used to generate all the assets in this game, including the ones that went unused. If you 100%'d that game, you used much, much more.
AI-assisted tools were used in the creation of some background images included in the game.
AI-assisted translation was also used for the English and Korean localization. The translated text was reviewed and adjusted during development.
There is no content generated by AI in real time during gameplay.
Some visual assets in Triad were created with the assistance of AI tools and further refined by the developer.
Triad is a solo-developed project, and these tools helped make its creation possible.
— Raph, developer of Triad
This game utilizes AI-assisted technologies in its development process. Generative AI was used to assist in writing code, refining localized text, and generating certain background ambient music and sound effects. Additionally, AI tools were used to help conceptualize and refine the unique "Shadow Puppetry" and "Ink-wash" artistic style assets to ensure a high-quality visual experience.
The game uses AI voice generation tools to create audio voiceovers for partial narrative texts and dialogue. All underlying text and story are originally created by the developer.
Some game artwork was generated with the assistance of AI, including monster sprites and a portion of achievement icons. These assets were then manually reviewed and edited to fit the game's pixel-art style. Store page descriptions were co-written with AI assistance.
Generative AI is used to assist with localization/translation. All localized text is reviewed and edited by the developer.
Generative AI is used to generate steam capsule image and asset. Act as placeholder
- Visual art: A portion of the pixel-art tiles and select character sprites were generated using PixelLab AI (pixellab.ai), then composited, edited, and integrated by the developer.
- Audio: Background music tracks were generated with AI music tools, then exported and sequenced into the game.
- Code & narrative writing: Game code and the in-game chronicler text (Brom's journal entries) were drafted with AI assistance (Anthropic's Claude) and edited by the developer.