Manuel Hüttl is CEO and shareholder of Milk & Honey Germany and Chairs the PRCA’s Innovation Working Group, which develops industry frameworks, standards, tools and guidance to support agencies as they deliver innovative, effective and responsible AI.
You know the story: a new technology gold rush unleashes a wealth of opportunities and a mess of problems; we enter a wild west where innovation outpaces ethical oversight and regulatory guardrails; before the regulatory posse catches up with the technology to bring peace and prosperity back to Dodge City.
AI is now deep into wild west territory: in the past few weeks alone we’ve heard of AI going rogue and a tech titan telling us all to saddle up and enjoy the ride. Yee har!
All that’s about to change. There’s a new sheriff in town — the EU — toting Article 50 of the EU AI Act to bring law, order and transparency to technology’s wild frontier.
Transparency is the keyword here, with important implications for agencies and brands.
AI transparency — from guidelines to guardrails
Until now, AI governance has been largely based on guidelines, such as the widely adopted 2023 ICCO Warsaw Principles. As with all guidelines, however, compliance relies on interpretation and self-regulation, which can lead to uneven application…or even wilful avoidance.
It’s a situation that has spurred the EU to action: as AI becomes ‘more human’, it’s increasingly important that humans understand when they engage with something that isn’t. None of us know definitively, for example, if the article we’ve just read is the outcome of opinion or algorithm.
Regulation brings the certainty of a level playing field. With the AI Act in full force, we must now (to quote directly from Article 50): “…ensure that AI systems intended to interact directly with natural persons are designed and developed in such a way that the natural persons concerned are informed that they are interacting with an AI system…”
See no evil, hear no evil, speak no evil
Article 50 means that from 2 August 2026, it’s a legal requirement to be clear when we’ve used AI — whether to wholly create or substantially amend a byline, blog, social post or anything else. This is particularly important around content dealing with public interest topics, classified as those covering: governance and law; society and welfare; and economy and science.
There is an exception: where an AI draft has been reviewed, edited and become the editorial responsibility of a human writer. I’d err on the side of caution here: because transparency is always best, and because I don’t know when something created by AI becomes substantively human — do you?
It doesn’t stop there. Beyond text, transparency now extends to images, video and audio — with the same provisions around public interest topics. The EU has created its own gallery of icons designed to ensure that creators and providers properly label AI content, ranging from: Basic; to AI Modified; to Fully AI Generated.
Bring the p-AI-n
The new EU sheriff carries some powerful weapons to enforce the law and bring order. Larger companies and SMEs that breach the Act face up to €15 million or 3% of revenue in penalties (whichever is higher for the former and whichever is lower for the latter). Should a company compound its error by supplying incorrect, incomplete or misleading information, it can also incur an additional fine of up to €7.5 million or 1% of global turnover.
Further, there’s nowhere for outlaws to hide. If a company thinks it’s beyond the EU’s jurisdiction, it will have to think again. Any AI system put on the market or deployed in the EU, and any AI content used within the EU, brings a company into scope.
That’s got to hurt.
EU law and AI order
Article 50 will cause a panicked stampede for compliance — but only among those who have sought to manipulate, distort and defraud.
For the rest of us, it won’t frighten the horses. Agencies committed to responsible AI have used accepted industry standards such as the ICCO’s 10 Warsaw Principles from day one; the very first of which states, “Disclosing the use of AI-generated content is vital, especially in the age of deep fakes.”
Article 50 simply provides the certainty and consistency of a legal foundation for the AI innovation we explore, the authentic work we do and the transparency we demonstrate every day.
Happy AI trails, partners!
