AI Usage at a Glance
Timeline
Sep 24, 2019
Creative GenPractice documented: Embark developed and open-sourced a texture synthesis tool in Rust that generates new textures from example images using a multiresolution stochastic algorithm, enabling artists to create variations from a single source texture.
Practice DocumentedView practice →Nov 20, 2020
ProductivityPractice documented: Embark built automated pipelines combining photogrammetry, Houdini-based procedural generation, and custom tools to convert real-world scan data into game-ready environments and characters, reducing asset creation time from weeks to days.
Practice DocumentedView practice →Feb 26, 2021
OtherPractice documented: Embark trains neural networks using reinforcement learning to control physics-based character locomotion, replacing traditional hand-animated state machines. Robot enemies learn to walk, balance, and react to damage emergently — if you shoot off a leg, the NPC attempts to rebalance on its own.
Practice DocumentedView practice →Mar 28, 2022
Data AnalysisPractice documented: Embark has developed prototype tools that use semantic and visual similarity algorithms to suggest relevant game assets to creators — for example, recognizing that a bottle is semantically similar to a bucket (both containers) or that a dead tree is visually similar to a lamp post.
Practice DocumentedView practice →Mar 28, 2022
ProductivityPractice documented: Embark Studios announced plans to build LLM-based tools for narrative design, quest building, lore creation, and player-facing story reactivity, with the work housed in its Machine Learning Research team in Stockholm. The studio's goal, stated in a job listing active as of early 2026, is to create new creative workflows for game writers and designers rather than automating existing ones. Research into language models for game characters was first described publicly by co-founder Magnus Nordin in March 2022.
Practice DocumentedView practice →Apr 4, 2023
ProductivityNew evidence: Embark's Erik Hallberg Discusses the Studio's Houdini-Powered Asset Processor
Evidence AddedView practice →Oct 30, 2023
Creative GenPractice documented: Embark uses AI text-to-speech technology trained on contracted human voice actors' recordings to generate in-game voice lines, including commentator dialogue, player character barks, and contextual callouts like ping system audio.
Practice DocumentedView practice →Dec 10, 2023
Creative GenNew evidence: Embark Studios: AI let devs "do more with less" when making The Finals
Evidence AddedView practice →Dec 13, 2023
RecommendationPractice documented: The Finals uses a skill-based matchmaking (SBMM) system that algorithmically evaluates player skill to create balanced matches, with independent skill ratings per game mode.
Practice DocumentedView practice →Dec 14, 2023
RecommendationNew evidence: SBMM in The Finals? This is what the developers say
Evidence AddedView practice →Mar 7, 2024
Data AnalysisNew evidence: Embark: A Technology-First Studio
Evidence AddedView practice →Apr 8, 2025
OtherPractice documented: The Finals integrates NVIDIA's DLSS (Deep Learning Super Sampling) technology, which uses AI neural networks to upscale lower-resolution frames in real-time, boosting performance by approximately 2x at 4K with ray tracing enabled.
Practice DocumentedView practice →Oct 23, 2025
ModerationPractice documented: Embark deploys machine learning models that analyze player behavior patterns (particularly mouse input, aiming, and recoil) to detect cheaters such as aimbot users, complementing traditional anti-cheat software like Easy Anti-Cheat and Denuvo.
Practice DocumentedView practice →Nov 4, 2025
Creative GenNew evidence: As ARC Raiders Takes Off, Developer Embark Defends Its Use of AI Tools
Evidence AddedView practice →Feb 20, 2026
ModerationNew evidence: As cheats improve, Arc Raiders dev Embark is fighting back in its squad FPS The Finals, and warns players to watch their software
Evidence AddedView practice →Mar 14, 2026
Creative GenNew evidence: Embark CEO says 'a real professional actor is better than AI' after the studio re-records some Arc Raiders dialog with real humans
Evidence AddedView practice →Recent activity
Embark developed and open-sourced a texture synthesis tool in Rust that generates new textures from example images using a multiresolution stochastic algorithm, enabling artists to create variations from a single source texture.
The texture-synthesis repository (~1,700 GitHub stars, now archived) implements a non-neural algorithmic approach where "every pixel asks 'if these are my neighbors, what is my color?'" based on example images. Tech artist Anastasia Opara and programmer Tomasz Stachowiak built it, generating approximately 3,000 images from ~150 style examples with 6 guide maps. Each image took ~20 seconds to generate. The tool supports guide maps, coordinate transforms, and multi-threading. While now archived, it demonstrates Embark's early investment in AI-adjacent creative tools and was used in their internal content pipeline.
Embark integrates with WebPurify, a third-party AI-powered content moderation service, for filtering user-generated text content in their games, as evidenced by their open-source API client library.
Embark's GitHub repository tame-webpurify is a Rust client for the WebPurify REST API, tagged with "content-moderation." WebPurify provides AI-driven profanity filtering, image moderation, and content safety services used across the gaming and social media industries. The existence of a purpose-built, open-sourced client library indicates production use of this content moderation service in Embark's multiplayer games, where player chat and user-generated content require filtering. Specific implementation details (which games, what content types) are not publicly documented beyond the repository's existence.
The Finals integrates NVIDIA's DLSS (Deep Learning Super Sampling) technology, which uses AI neural networks to upscale lower-resolution frames in real-time, boosting performance by approximately 2x at 4K with ray tracing enabled.
The Finals launched in December 2023 with NVIDIA DLSS 3 support (AI upscaling + AI frame generation), DLAA (AI-based anti-aliasing), NVIDIA Reflex (latency reduction), and hardware-accelerated ray tracing via RTXGI for dynamic global illumination. The AI upscaling is particularly important for The Finals because its destructible environments require real-time lighting recalculation. In April 2025, Embark added DLSS 4 Multi Frame Generation support for RTX 50 Series GPUs. While DLSS is NVIDIA's technology (not developed by Embark), its integration is a deliberate choice to deliver AI-enhanced rendering to players. NVIDIA released dedicated Game Ready Drivers for The Finals' launch.
All practices by category
Embark developed and open-sourced a texture synthesis tool in Rust that generates new textures from example images using a multiresolution stochastic algorithm, enabling artists to create variations from a single source texture.
Embark uses AI text-to-speech technology trained on contracted human voice actors' recordings to generate in-game voice lines, including commentator dialogue, player character barks, and contextual callouts like ping system audio.
Embark integrates with WebPurify, a third-party AI-powered content moderation service, for filtering user-generated text content in their games, as evidenced by their open-source API client library.
Embark deploys machine learning models that analyze player behavior patterns (particularly mouse input, aiming, and recoil) to detect cheaters such as aimbot users, complementing traditional anti-cheat software like Easy Anti-Cheat and Denuvo.
The Finals integrates NVIDIA's DLSS (Deep Learning Super Sampling) technology, which uses AI neural networks to upscale lower-resolution frames in real-time, boosting performance by approximately 2x at 4K with ray tracing enabled.
Embark trains neural networks using reinforcement learning to control physics-based character locomotion, replacing traditional hand-animated state machines. Robot enemies learn to walk, balance, and react to damage emergently — if you shoot off a leg, the NPC attempts to rebalance on its own.
Embark Studios announced plans to build LLM-based tools for narrative design, quest building, lore creation, and player-facing story reactivity, with the work housed in its Machine Learning Research team in Stockholm. The studio's goal, stated in a job listing active as of early 2026, is to create new creative workflows for game writers and designers rather than automating existing ones. Research into language models for game characters was first described publicly by co-founder Magnus Nordin in March 2022.
Embark built automated pipelines combining photogrammetry, Houdini-based procedural generation, and custom tools to convert real-world scan data into game-ready environments and characters, reducing asset creation time from weeks to days.
Related Companies
Trace Foundation
Krafton
Activision Blizzard
Walmart
Electronic Arts (EA)
Have evidence about Embark Studios's AI practices? Submit a report.
Submit a report →AI Trace is free and nonprofit. Support our work