Why new EU deepfake legislation means games need to be careful when featuring real-world people and locations

The EU AI Act’s deepfake transparency rules are now a real production concern for game teams, with disclosure obligations applying from 2 August 2026. The definition is broader than the usual “fake video” shorthand: it covers AI-generated or AI-manipulated audio, video, and images that resemble real or plausibly real people, places, objects, entities, or events and could be taken as authentic.

Recent European Commission guidance gives studios some welcome breathing room. Common game uses such as AI voice work for fictional characters, fantasy environments, and background or post-processing effects are unlikely to count as deepfakes when there’s no deception about identity or reality. That matters because a lot of routine AI-assisted content in games sits comfortably in the creative-fictional lane.

The grey area starts when games lean on realism. Recreating a celebrity voice, depicting a real athlete or historical figure, or generating photorealistic promotional material around real-world events can trigger the label requirement. The same goes for marketing assets outside the game itself: trailers, billboards, influencer campaigns, and social posts can all fall under the rule if they’re distributed in the EU.

The practical takeaway is that studios need a content policy, not just a legal memo. Teams should be trained to ask whether the content depicts something real or plausibly real, whether it looks highly realistic, whether audiences could believe it’s authentic, and whether it will reach the EU. For most studios, the challenge is less about banning AI and more about...