AI Act Mandates Disclosure for Deepfake Content

Under the new EU AI Act, creators, advertisers, and political campaigns must explicitly label all AI-generated or manipulated media to prevent public deception, facing heavy fines for non-compliance.

The New Transparency Requirement

Deepfakes are no longer just a reputation or technological concern. The AI Act mandates that any audio, image, or video material generated or significantly altered by AI must be disclosed as synthetic if it could be mistaken for authentic content. This obligation spans across advertising, political communication, and digital information.

Violations of these transparency rules can result in severe financial penalties reaching up to 15 million euros or 3 percent of a company’s total worldwide annual turnover.

Defining the Scope of Synthetic Content

The regulation adopts a broad definition of deepfakes, covering any AI-manipulated material that mimics real persons, objects, places, or events. This includes not only face-swaps but also synthetic voices, manufactured statements, and staged photos of events that never occurred.

Transparency must be clear and provided no later than the recipient’s first contact with the material. Burying disclosures in terms and conditions or invisible footers is not a compliant solution.

Political Campaigns Under Scrutiny

Political entities using AI must disclose the synthetic nature of their materials to prevent misinformation. While the AI Act does not ban generative AI in politics, it prohibits misleading voters regarding the authenticity of content. Satirical or artistic works remain permitted, provided the disclosure does not impede the audience’s ability to enjoy the piece.

Marketing and Professional Responsibilities

Agencies and influencers share responsibility for labeling content alongside the developers of AI tools. In advertising, this is critical when AI mimics experts, customers, or celebrities to suggest authentic endorsements. Companies are encouraged to separate creative content from realistic depictions and maintain thorough documentation of their AI usage.

Enforcement and Penalties

The European Commission has provided a code of practice to help entities fulfill these obligations, though the AI Act itself is the binding legal framework. When determining fines for non-compliance, authorities will consider factors such as the scale of the deception, the entity’s size, and any financial benefits gained from the violation.