AI Generated or AI Modified? Understanding the EU’s New Transparency Rules
From 2 August 2026, new transparency requirements under the EU AI Act apply to certain AI-generated and AI-manipulated content in the European Union. For businesses, this raises an increasingly important question: when does the use of AI need to be disclosed, and what does transparency look like in practice?
Under Article 50 of the EU AI Act, the answer depends on both the type of content and the organization’s role. Providers of generative AI systems face requirements to make certain AI-generated outputs detectable in a machine-readable format, while businesses using AI systems – known as deployers under the Act – face disclosure obligations in specific circumstances, particularly for deepfakes and certain AI-generated or manipulated text concerning matters of public interest.
For many organizations, this may sound like another compliance exercise. In reality, it represents something much bigger. The introduction of official EU labels marks an important shift in how businesses communicate their use of artificial intelligence, placing transparency, accountability, and trust at the centre of digital content. Companies that prepare early won’t just be ready for the regulation – they will be better positioned to build credibility with customers, partners, and stakeholders in an increasingly AI-driven world.
This article explores what the new EU AI transparency requirements mean in practice, explains the distinction between AI-generated and AI-modified content, and outlines how businesses can turn transparency from a regulatory consideration into a strategic advantage.
What Article 50 Actually Requires
The EU AI Act (Regulation (EU) 2024/1689) is the world’s first comprehensive legal framework for regulating artificial intelligence, and it rolls out in stages. Article 50 governs transparency and creates distinct duties depending on your organization’s role.
Providers (companies that develop and place certain AI systems on the market) must ensure that the outputs of AI systems generating synthetic audio, image, video or text content are marked in a machine-readable format and detectable as artificially generated or manipulated. These technical solutions must be effective, interoperable, robust and reliable as far as technically feasible.
Deployers (organizations using AI systems) face a different and narrower set of disclosure obligations. In particular, deployers must disclose when image, audio or video content constitutes a deepfake. They must also disclose when AI-generated or manipulated text is published for the purpose of informing the public on matters of public interest, unless the content has undergone human review or editorial control and a person or organisation holds editorial responsibility for its publication.
This distinction is important. Article 50 does not impose a general requirement for every business to visibly label every piece of AI-generated or AI-modified marketing content.
Two dates are particularly relevant:
2 August 2026: Article 50’s transparency obligations become applicable. This includes the relevant disclosure requirements for deployers and technical marking requirements for providers.
2 December 2026: providers of AI systems already placed on the market before 2 August 2026 have additional time to comply with the technical marking obligations under Article 50(2).
The rules are also not generally retroactive. AI-generated or manipulated outputs that were generated and already made available before 2 August 2026 do not need to be retroactively marked or labelled. However, organisations publishing content after that date should assess whether the relevant Article 50 obligations apply.
AI Generated vs. AI Modified: Why the Distinction Matters
This is where the real strategic challenge begins. The EU AI Act itself refers broadly to content that is “artificially generated or manipulated” rather than establishing AI Generated and AI Modified as two separate legal classifications.
However, distinguishing between content created primarily by AI and human-created content subsequently modified with AI can still be useful from a governance, transparency, and content-management perspective.
AI Generated content can generally be understood as content originally created by an AI system based on a prompt. A product image generated from a text prompt using an AI image generator, for example, would clearly have an AI origin even if a human later makes minor edits.
AI Modified content can describe fundamentally human-created content that has subsequently been altered using AI – for example, background adaptation, image enhancement, generative expansion, or other AI-assisted refinements.
The boundary isn’t always clean. Different AI tools can play very different roles in the creative process, ranging from minor technical assistance to substantial generation or manipulation. Businesses should therefore avoid treating these informal categories as definitive legal classifications.
Instead, the strategic takeaway is clear: businesses should establish a documented and consistent approach to recording how AI is used at the point of creation. Beyond supporting compliance where disclosure requirements apply, this creates clearer approval workflows, reduces legal uncertainty, and ensures greater consistency across marketing, creative, and communications teams.
| Transparency Category | Example | Practical Example |
|---|---|---|
| AI Generated | Content created primarily by an AI system based on a prompt. This is a useful internal classification rather than a separate legal category established by Article 50. | A product image created in Midjourney or ChatGPT Image from a text prompt. |
| AI Modified | Human-created content that has subsequently been materially altered or enhanced using AI. Again, this can be useful for internal governance but should not be treated as a distinct legal category under Article 50. | A photograph where AI adapts the background, improves elements of the image, or expands the scene. |
| AI Disclosure | A disclosure used where required by the EU AI Act or adopted voluntarily as part of an organisation’s broader transparency policy. | A business clearly informing viewers that qualifying content has been artificially generated or manipulated. |
Who This Applies To
The new transparency requirements extend beyond major AI providers, but the obligations differ depending on the role an organisation plays.
Providers of generative AI systems are responsible for the technical marking requirements applicable to their systems. Businesses using those systems as deployers have disclosure obligations in more specific circumstances.
Article 50 places particular emphasis on deepfakes and AI-generated or manipulated text relating to matters of public interest. For text published to inform the public on matters of public interest, the disclosure obligation does not apply where the AI-generated content has undergone a process of human review or editorial control and where a natural or legal person holds editorial responsibility for its publication.
There are also specific rules for evidently artistic, creative, satirical, fictional or analogous works or programmes. In these cases, the transparency obligation is limited so that disclosure does not hamper the display or enjoyment of the work.
This means that an ordinary business does not automatically need to place a visible AI label on every marketing image, social media post, blog article, or other asset simply because AI contributed to its creation.
For most organisations, however, the key takeaway remains simple: don’t focus solely on the minimum legal requirement. As AI becomes an increasingly common part of content creation, adopting transparent internal standards helps strengthen governance, reduce uncertainty, and build trust with customers and stakeholders.
Understanding the New EU AI Labels
To support implementation of the new transparency requirements, the European Commission’s AI Office has developed a Code of Practice on Transparency of AI-Generated Content, together with visual tools intended to make AI disclosures more consistent and recognizable.
The European Commission has introduced optional icons that organizations can use to communicate the involvement of AI in content. These visual labels can provide businesses with a consistent way of approaching AI transparency, but they should not be confused with mandatory legal classifications established by the AI Act itself.
While the official labels offer a common visual standard, their use is voluntary. What is mandatory is compliance with the disclosure obligations set out in Article 50 wherever those obligations apply. In other words, businesses are not required to use the Commission’s icons specifically, nor does every piece of AI-generated or AI-modified business content automatically require a visible label.
For businesses, the value of the new labels extends beyond regulatory compliance. Applying a consistent approach to AI disclosures across websites, social media, marketing campaigns, and other digital assets helps establish clear internal standards while reinforcing transparency and trust. As AI becomes an increasingly integral part of business communication, a consistent labeling strategy is likely to become as much a branding decision as it is a compliance consideration.
The Cost of Getting This Wrong
The financial penalties under the EU AI Act are substantial, with fines for infringements of certain obligations under the Act reaching up to EUR 15 million or 3% of worldwide annual turnover. The Act also provides for proportionate treatment, including specific considerations relating to smaller businesses.
Yet for most businesses, the greater risk lies elsewhere.
As AI becomes a standard part of business communication, transparency is rapidly evolving into an industry expectation. Companies that establish clear AI governance and consistent labeling practices today will be better positioned to build trust, reduce compliance risks, and adapt to future regulatory developments.
Rather than waiting for enforcement, businesses should use this period to embed transparency into their everyday content and marketing workflows.
AI Transparency and the Future of AI Search
The introduction of the EU’s AI transparency requirements reflects a broader shift in the digital landscape. As generative AI becomes an increasingly common part of content creation, transparency is evolving from a regulatory requirement into a business expectation.
This shift is particularly relevant as AI-powered search engines and answer engines continue to change how people discover information online. Rather than simply listing links, these systems increasingly analyse, summarise, and recommend content based on its relevance, quality, and credibility. In that environment, trust becomes more important than ever.
While there is currently no evidence that AI labels directly influence search visibility, they represent a broader movement towards transparent and responsible digital communication. Businesses that clearly communicate how AI is used across their content demonstrate stronger governance, reinforce brand credibility, and help build confidence among customers, partners, and other stakeholders.
For organizations investing in AI Search and Generative Engine Optimization (GEO), transparency should not be viewed as a limitation. Instead, it should form part of a broader strategy focused on creating trustworthy, high-quality content that is valuable for both people and AI-powered discovery systems.