A marketing agency called BrandBoost ran an advertising campaign featuring a video of a well-known scientist appearing to endorse a health supplement. The video was AI-generated — the scientist never agreed to appear. When challenged, BrandBoost's compliance bot replied: "The video was produced using standard digital production techniques. Our creative team holds all appropriate content licences. We do not consider this content to fall under any disclosure requirement."

The Situation

BrandBoost used an AI video synthesis model to generate footage of a real person endorsing a product. The person's likeness, voice, and mannerisms were synthesised — they said nothing of the sort. The video ran across social media platforms with no label, no disclosure, and no indication that it was AI-generated.

The EU AI Act introduced explicit mandatory disclosure rules for exactly this scenario. "Digital production techniques" is not a loophole — it is the definition of what the law covers.

Key Law

Arguments That Strengthen Your Case

  1. Establish identifiability first — the AI Act deepfake provision applies when the content noticeably resembles a specific, real, identifiable person. Confirm the subject is a named, recognisable individual. If so, Article 50(4) applies directly.
  2. Cite Article 50(4) directly — disclosure of AI-generated deepfake content is mandatory, not optional. "Standard digital production techniques" does not create an exemption from this obligation.
  3. Challenge the "licences" claim — content licences govern copyright, not the AI Act disclosure obligation. They are different legal frameworks. Holding a content licence does not satisfy the transparency requirement under Article 50(4).
  4. Invoke the Unfair Commercial Practices Directive — using a false endorsement (even AI-generated) of a real person constitutes a misleading commercial practice, creating a second, independent legal basis for the complaint.

About This Case

Must companies label AI-generated videos and images?

Yes. EU AI Act Article 50(4) requires persons who deploy AI systems to generate or manipulate image, audio, or video content to disclose that the content has been artificially generated or manipulated. This applies to deepfakes in advertising, social media, and promotional material. The disclosure must be machine-readable and visible to the end recipient.

What if the company says the video is "digitally enhanced" rather than AI-generated?

The EU AI Act applies to content that noticeably resembles existing persons. If an AI model generated or synthesised video content of a real person, disclosure is required regardless of what the company calls the production method. "Digital enhancement" is not a meaningful distinction under the law.

How do I practice this case?

Play this case free at fixai.dev — no account required.

Related Cases

BrandBoost ran an AI-generated video of a recognisable scientist endorsing a supplement — no consent, no disclosure. BRANDBOT-AI says "content licences held." Cite Article 50(4) and force the campaign to be suspended.

Play This Case →
← All Cases