In July, the EU's Digital Omnibus on AI entered into force, and many headlines summed it up as "the EU delayed the AI Act." That is only half true. The obligations for most high-risk AI systems were pushed back to December 2027. But the transparency rules in Article 50 were not delayed — they have applied since August 2, 2026.
If your organization uses AI to talk to customers or to produce content, this part of the law is already relevant.
What Moved and What Didn't
Delayed: Most obligations for high-risk AI systems listed in Annex III of the Act — the systems used in areas like hiring, credit and education — now apply from December 2, 2027 instead of August 2026.
Not delayed: The Article 50 transparency obligations applied on August 2, 2026, as originally planned. The AI Office's enforcement powers over providers of general-purpose AI models also became active.
Extra time for one piece: Systems already on the market before August 2 get a transition period, until December 2, 2026, for the technical requirement to mark AI-generated content in a machine-readable way.
What Article 50 Asks For
In broad terms, Article 50 is about making sure people know when AI is involved:
- People should be told when they are interacting with an AI system, such as a customer-service chatbot, unless it is obvious from the context
- AI-generated audio, images, video and text should be marked so it can be detected as artificially generated
- Deepfakes — realistic content showing real people, places or events — must be disclosed as AI-generated
- AI-generated text published to inform the public on matters of public interest must be disclosed, unless it has gone through human review and someone takes editorial responsibility for it
Some of these duties fall on the companies that build AI systems, others on the organizations that use them. Which apply to you depends on your role and use case, so check the specifics with legal counsel — this article is a summary, not legal advice.
Why This Is a Workflow Question, Not Just a Legal One
The practical challenge is knowing where AI was used. If a marketing team, a support desk and a policy unit all use AI in different ways, nobody may have a clear picture of which content was AI-generated and which was reviewed by a person.
Teams that already work in a structured way have a head start:
- Known prompts: When the instructions given to AI are written down rather than improvised, it's easy to see what was generated and why.
- Human review built in: If drafts always pass through an editor, the human-review exception for public-interest text becomes much easier to show.
- Clear disclosure points: Customer-facing chatbots should identify themselves from the first message, not in the fine print.
Getting Ahead of It
You don't need a large compliance program to start. Make a list of where your team uses AI, write down the standard prompts for recurring tasks, and decide which outputs need a human sign-off before they go out.
Structured prompts help with all three. IacuWise turns ad hoc requests into clear, reusable prompts — so your team gets better results today and has a clearer record of how AI is being used.