Disclosing AI-made creative: the rules so far

A mobile game creative made partly or wholly with generative AI now sits under two separate kinds of rule, one platform-level and one regulatory, and a studio that has only checked one of them isn't actually compliant. Meta and TikTok have each built their own disclosure mechanism for AI-generated or AI-altered content inside ad creative, distinct from the wider transparency obligations moving through implementation under the EU's AI Act and, in practice, more specific than them. A creative team briefing a new concept needs to know which rule applies to which asset. The two frameworks trigger on different things.

What the platforms already require

Meta's ad policy requires advertisers to disclose when a creative has been digitally created or altered in ways that would materially change how a viewer perceives the product. Meta built the rule originally for political and social issue ads; it then extended it to commercial creative that uses synthetic voices or generated likenesses, and to substantially AI-altered gameplay footage. TikTok runs a comparable disclosure requirement for AI-generated content in ads, with its own labelling mechanism inside Ads Manager. Neither platform wants a label on every use of AI in production. A fully AI-assisted storyboard doesn't trigger disclosure on either platform, and nor does an AI-upscaled asset with no altered representation of the product, but both require it once the output could reasonably mislead a viewer about what the game actually looks or plays like.

That distinction, between AI used as a production tool and AI used to alter what an ad appears to show, is the one that matters operationally. A studio using generative tools to produce backgrounds or voiceover, or to spin iteration variants off an otherwise real gameplay capture, is on safe ground without a label. A studio using AI to generate footage that corresponds to no real build, dressed up to look like gameplay, is exactly the case both platforms built their disclosure rules to catch; it's also the case most likely to draw regulatory attention regardless of platform policy.

Where the EU AI Act adds a separate layer

The EU AI Act includes a transparency obligation that AI-generated or manipulated content must carry a marking saying so. The provision initially targeted deepfakes and synthetic media more broadly rather than advertising specifically, but its wording is broad enough to reach advertising content distributed inside the EU as its various articles come into force on their staggered timeline. Studios advertising into EU markets should treat the Act's transparency requirement as a compliance floor that sits underneath whatever a given ad platform already demands, not as a replacement for it. A platform's disclosure label doesn't automatically satisfy a statutory transparency obligation; the two carry different enforcement mechanisms and different penalties.

What the UK's regulator already covers without a new AI-specific rule

The UK's Advertising Standards Authority hasn't needed a bespoke AI rule to act on misleading game ads, because its existing framework against misleading advertising already applies regardless of how a misleading image came to exist. An ad showing gameplay that doesn't exist in the shipped product breaches existing rules whether the false footage came out of an animator's software or a staging rig or a generative model. The tool doesn't matter. Studios that assume AI-specific regulation is the only relevant compliance risk are missing the older, broader rule that already covers the actual harm regulators care about: a misleading impression of the product, independent of the tool used to create it.

The practical checklist for a studio right now

What a studio must label today is any commercial creative where AI has materially altered the viewer's impression of what the game looks or plays like, checked against whichever platform is running the ad. What is currently only good practice, rather than a hard requirement, is broader disclosure of AI use in production tooling that doesn't change what the ad appears to show. Studios expecting the regulatory framework's scope to widen over time would be sensible to start that habit now rather than after it becomes mandatory.

The safest working rule for a creative team briefing a new AI-assisted concept is to ask what a player would assume about the game after watching the ad, and to check that assumption against the actual build before worrying about which platform's label applies.

Every disclosure rule covered here sits downstream of that one question.

Related archive reading

These articles provide related context and remain subject to their stated review status.

• AI-Generated Ad Creative: Where the Quality Bar Really Sits
• AI storyboarding is faster. Is it better? A side-by-side