How Deep is Your Fake? A 3-Minute-Guide on Labelling Obligations under the EU AI Act

2026-07-15 · Source: Privacy World · Field: Legal & Regulatory — Compliance & Risk Management, Regulatory Affairs & Government Relations · Depth: Intermediate, short

Summary

The EU AI Act mandates disclosure for AI-generated or modified content, with enforcement beginning August 2, 2026, carrying fines up to EUR 15 million or 3% of global annual turnover. Labelling obligations primarily apply to AI-generated/manipulated texts informing the public on matters of public interest, such as corporate reports with investor information, and to "deep fakes." The Act defines deep fakes broadly in Article 3(60) as AI-generated or manipulated image, audio, or video content resembling existing persons, objects, places, entities, or events that would falsely appear authentic. This broad scope impacts product advertising, requiring companies to assess if content might deceive an "average person," particularly for sensitive groups. Practical guidance from the EU AI Office's Code of Practice suggests using standardized icons, ensuring clear visibility, embedding disclosures in metadata, and providing audio labels for relevant content. Content undergoing human review is exempt.

Key takeaway

For manufacturers, retailers, and advertising creators operating within the EU, you must proactively establish robust internal guidelines and procedures for AI transparency before August 2, 2026. Your teams should identify and clearly label all AI-generated or modified content, especially deep fakes in advertising, to avoid substantial fines and potential unfair competition claims. Regularly monitor sector developments to ensure your compliance measures remain current and effective.

Key insights

The EU AI Act mandates broad labelling for AI-generated/modified content, particularly deep fakes in advertising, with significant compliance implications.

Principles

• AI Act labelling applies to public interest texts and deep fakes.
• The deep fake definition extends beyond persons to objects, places, entities, and events.
• Authenticity assessment uses a hypothetical "average person" standard, considering target audience literacy.

Method

Implement clear, noticeable labels prominently within visible content, embed in metadata, and use AI watermarks; audio content requires clear audible disclosure, and video needs continuous or regular interval labelling.

In practice

• Utilize freely available standardized EU icons for labelling.
• Consider distinguishing "AI Generated" from "AI Modified" as encouraged.
• Establish internal guidelines and procedures for AI transparency and compliance updates.

Topics

• EU AI Act
• AI Content Labelling
• Deep Fakes
• Advertising Compliance
• Regulatory Compliance

Best for: Legal Professional, Marketing Professional, Director of AI/ML

Related on AIssential

• AI transparency requirements in marketing — The EU AI Act mandates clear labeling for AI-generated or manipulated content in marketing to prevent public deception.
• EU publishes its AI content labelling playbook ahead of the AI Act’s August deadline — The EU's AI Act mandates clear labelling for AI-generated content and interactions to ensure transparency by August 2, 2026.
• Commission publishes guidelines on transparency obligations for providers and deployers of certain AI systems — The EU AI Act mandates transparency for AI systems to prevent user deception and manipulation.
• AI Transparency Deadline Approaching — The EU AI Act mandates strict transparency and disclosure for AI systems, particularly concerning human interaction, generated content, and biometric data.
• Scientists Asked AI to Impersonate 112 Public Figures. What Happened Next Is a ‘Dire’ Warning — AI-generated impersonations of public figures are perceived as more authentic and coherent than real human responses, posing a "dire" deception risk.
• EU AI Act Article 50: Transparency Rules for Businesses — The EU AI Act's Article 50 mandates transparency for diverse AI systems, requiring users to know when they interact with or consume AI-generated content.

See Counsel's argued verdicts on the open AI decisions leaders are weighing →

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Privacy World.