Last Updated on 17. August 2026 by Kraus
AI Act / EU AI Regulation – Transparency & Labelling Requirements for AI Content
Introduction
The AI Act, also commonly called the EU AI Act or EU AI Regulation, establishes clear rules from 2026 for the use of artificial intelligence. This page explains in plain language what the regulation means for businesses, creators and website operators and how AI is used responsibly on delavo.de.
Note: This page provides general, practical guidance and does not constitute legal advice. For a legally binding assessment of your individual situation, we recommend consulting a lawyer specialising in IT and media law.
What is the AI Act (EU AI Regulation)?
The AI Act, officially the EU AI Regulation, is the European Union’s new legal framework for regulating artificial intelligence. It sets requirements for how AI systems must be developed, deployed and monitored to ensure safety, transparency and consumer protection. The regulation classifies AI systems by risk level and defines specific obligations for providers, companies and public authorities. For content creators and website operators the “limited risk” category is particularly relevant because it includes transparency and labelling duties. Non-compliance with the transparency obligations can result in fines of up to €15 million or 3% of global annual turnover – one reason delavo.de takes this particularly seriously.
Labelling and disclosure requirements under the AI Act / EU AI Regulation
The AI Act requires that AI‑generated content be made clearly identifiable when users would otherwise not notice it was produced by AI. This covers AI images, AI videos, synthetic voices and fully automated text generation. Websites must clearly indicate when content was produced without human review or when users are interacting with an AI system. If content has been editorially reviewed and a human takes responsibility for the final output, labelling as purely AI‑generated is not required. The labelling obligation for text applies only to content that informs the public on matters of public interest (e.g. political, economic or societal topics) in the first place — plain product reviews such as those on delavo.de generally fall outside this obligation altogether.
Providers vs. deployers: who is responsible for what?
The AI Act distinguishes between providers (the makers of AI systems, e.g. ElevenLabs or Synthesia) and deployers (businesses and individuals who use those systems, such as delavo.de). Providers carry their own obligations, including machine‑readable marking of generated content. As a deployer, delavo.de additionally carries responsibility for visible, human‑readable labelling towards its own readers.
How AI is used on delavo.de
I use AI tools to make workflows more efficient while following a clear, responsible process.
Research
All content is based on manual research carried out by the author. Features, pricing, specifics and practical suitability of tools are checked personally.
AI assistance for text
AI supports drafting, structuring and optimizing texts. Every text is reviewed, edited and approved by a human author who remains fully responsible for the content.
One exception are product pricing tables: initial prices are verified and set by the author, while subsequent price updates are performed automatically by AI. These tables are labelled accordingly — on German pages with “(Preis automatisch aktualisiert durch KI)” and on English pages with “(Price automatically updated by AI)”.
AI images
Images generated or edited by AI are clearly labelled. On German pages the label reads “Bild KI‑bearbeitet”, on English pages “Image AI‑edited”. This label is applied deliberately even in cases where the AI Act would not strictly require it – for example, light image edits without deepfake character. This higher transparency standard is a voluntary editorial choice.
Technical assistance
AI helps with tables, comparisons, outlines and formatting. Final decisions and responsibility always remain with the author.
With this approach delavo.de meets the AI Act’s requirements and transparently shows how AI is used.
What the AI Act means for users
The AI Act ensures AI content is traceable and transparent. For most users this translates into simple, practical rules that are easy to implement.
Labelling requirement for AI images
Images that are fully or predominantly generated by AI must be clearly labelled. A short caption beneath the image is sufficient.
Labelling requirement for AI videos
AI videos must be identified when they are artificially generated or heavily altered. The notice can appear under the video or in the description.
Labelling requirement for AI audio
Synthetic voices must be labelled so listeners understand the voice is not real.
AI text
AI‑generated text does not need a label if it has been reviewed, edited and approved by a human. Only fully automated, unreviewed texts require a disclosure.
This means that using AI as an editorial aid and publishing content after human review is already compliant with the AI Act.
What does not need to be labelled
The following AI‑assistance is not subject to labelling requirements:
- Spelling correction
- Grammar improvement
- Style enhancement
- Structuring
- Summaries
- Translations
- Technical assistance such as tables or formatting
These forms of AI use are considered normal digital assistance.
Practical guide for safe use of AI tools
This compact overview helps you use AI content safely and transparently.
AI text
If texts are reviewed and approved by a human, no labelling is required.
AI images
Always label AI images.
AI videos
Always label AI videos.
AI audio
Always label AI audio.
Chatbots
Users should be able to recognise they are interacting with an AI system.
Documentation
A short internal note is sufficient to document AI usage and make it traceable.
Practical examples – matching the tool tests on delavo.de
Many tools reviewed on delavo.de use AI to generate content, voices or web experiences. The examples below show when the AI Act requires labelling and when it does not:
- ElevenLabs, Murf AI, Hume AI (AI voices & audio) → Labelling required: synthetic voices and AI‑generated voiceovers must be clearly identified for listeners.
- Synthesia (AI videos) → Labelling required: videos created with artificial avatars and synthetic voices must be clearly labelled.
- Riverflow AI, AdCreative.ai (AI images & ads) → Labelling required: photorealistic images or graphics fully generated by AI require a visible notice (e.g., under the image).
- Chatbase, Tidio (AI chatbots) → User notice required: website visitors must be informed when a chat is powered by AI.
- Rytr, QuillBot, Consensus (AI text & research) → No labelling required: when generated text is reviewed and edited by a human before publication, the labelling exception applies.
- Gamma, Prezi AI, Beautiful.ai, Unbounce AI, Instapage (AI presentations & landing pages) → No general labelling required: layout, structure and draft text are digital assistance. If fully automated AI images or videos are embedded, those elements must be labelled.
- Wispr Flow, Browse AI, Netlify, ActiveCampaign, Atria (productivity & automation) → No labelling required: dictation, web scraping, hosting, marketing automation and analytics are technical support and not subject to content labelling rules.
These examples show that most tools featured on delavo.de can still be used easily under the AI Act with minimal effort.
Current status: Digital Omnibus (2026)
In May 2026 the EU adopted amendments to the AI Act known as the “Digital Omnibus”. The general transparency obligations under Article 50 still apply from 2 August 2026. However, the requirement for machine‑readable marking (e.g. watermarking in metadata) was postponed to 2 December 2026. As the regulation is still evolving, this status may change — this page will be updated as relevant amendments occur.
Conclusion
The AI Act creates clear, easy‑to‑understand rules for handling AI content. When AI is used responsibly, content is reviewed and transparently labelled, the requirements are already met.
On delavo.de AI is used in a way that makes content creation traceable and understandable while giving readers practical guidance on how to use AI tools safely and professionally.
FAQ on the AI Act
Do I have to label AI texts?
AI texts must be labelled only if they are fully automated and published without human review. If you check, edit and take responsibility for the content, no labelling is required. The obligation also only applies to text that informs the public on matters of public interest — such as political, economic or societal topics.
Do AI images always need to be labelled?
Yes. Images generated or significantly altered by AI must be clearly labelled. A short caption under the image is sufficient.
Do the labelling rules apply to AI videos?
Yes. AI videos must be identified when they are artificially generated or heavily modified. The notice can be placed under the video or in the description.
Do synthetic voices need to be labelled?
Yes. AI‑generated voices must be clearly labelled so it is obvious they are not real human voices.
Do chatbots need to be labelled?
Yes. Users should be able to recognise they are interacting with an AI system. A short notice in the chat window or the chat introduction is sufficient.
Do AI‑assisted corrections like spelling or grammar need to be labelled?
No. Spelling correction, grammar optimisation and style improvements are considered normal digital assistance and do not require labelling.
Do summaries or translations by AI need to be labelled?
No. As long as you review and take responsibility for the content, summaries, translations and structural assistance do not require labelling.
How do I label AI content correctly on my website?
A short, visible notice such as “This image was created with AI” or “This video contains AI‑generated content” is sufficient. The label should be clear and easy to understand.
Is internal documentation about AI usage enough?
Yes. A brief internal note about which AI tools you use and how content is reviewed is sufficient. Extensive documentation is not necessary for normal content production.
What does the AI Act mean for practical use of AI tools?
The AI Act mainly requires transparency and responsible handling of AI content. If you review content, label it clearly and work in a traceable way, you can use AI tools safely without extra effort.