Are you and your team ready to comply with the European Artificial Intelligence Regulation, known as the AI Act? Read on to understand what the AI Act is for, when it comes into force, and what teams need to prepare before publishing AI-assisted content.
Artificial intelligence is part of our daily lives in routine tasks that we do not consider to be ‘AI projects’. Summarising a report, translating a note, adapting a text, reviewing the tone or generating an accompanying image are all tasks that these tools can streamline.
However, when this content is intended for publication and dissemination, the situation changes. It is the impact that these changes may have on society that has led the European Union (EU) to establish common rules for the development and use of AI systems through the AI Act.
The regulation is aimed at those who develop or market these technologies, but it also incorporates transparency and obligations, under Article 50, for organisations or individuals who use AI in their professional activities.
If you’re interested in communications or work in the sector, I invite you to read on to understand how this will change your day-to-day work: what preparation you and your team need, and when you must inform the public about an interaction or content generated or manipulated using AI.
To determine whether your company or business is affected by this regulation, the AI Act clearly distinguishes between a ‘provider’ and a ‘deployer’ in Article 3 of the AI Act.
According to this distinction, the provider is the party that designs, develops or commissions the development of an AI system, offers it under their own name or brand, and guarantees its technical functioning.
The deployer is the person, company, institution or body – such as communications teams – that uses that system under its authority during a professional activity; it bears editorial responsibility and ensures the transparent use of the tools and content it generates.
Therefore, organisations that have not developed the technology but use it to produce content or handle enquiries also have obligations regarding AI.
If you use AI in chatbots, or to generate images, audio, videos and text in your day-to-day work, the AI Act affects you in two ways:
The obligation regarding AI literacy has been in force since 2 February 2025. Every organisation must adopt measures tailored to its team’s experience, the context in which it uses AI, and the people who may be affected.
For communications teams, this requires far more than simply learning how to give instructions to a tool. It means developing skills in fact-checking, editorial review, reputational awareness, data protection, practical transparency and accountability.
Learning to use the tools can improve productivity; developing sound judgement also helps to safeguard trust.
Transparency is one of the obligations set out for the use of AI in Article 50. Of particular interest to communicators, it is worth distinguishing between four situations.
This exception does not mean that review is merely a formality. Someone must verify facts, preserve context, identify omissions and approve the final version.
Therefore, not all AI-assisted content should be labelled in the same way. Correcting a typographical error is not the same as publishing a realistic video of a person or a text about a public decision. The useful question is: does knowing that AI has been used change the way the public interprets this piece?
In June 2026, the Commission published the Code of Good Practice on Transparency of AI-Generated Content. Adherence to it is voluntary, but the obligations under Article 50 are legally binding. The code can serve as a reference for translating these into consistent internal criteria.
To incorporate these obligations into day-to-day work, it is advisable to establish five criteria:
From the regulation’s entry into force on 1 August 2024, the AI Act will be implemented progressively. In practice, the regulations stipulate that companies and organisations must be taking steps to promote AI literacy from 2 February 2025; therefore, from that date onwards, people working with these tools should possess the appropriate knowledge. Finally, 2 August 2026 is the general date for the implementation of the transparency obligations set out in Article 50, as explained above.
The regulations set out obligations and limits but cannot anticipate every editorial decision. There are risks that each organisation must manage according to its own judgement:
The AI Act is forcing many organisations to take a closer look at how they use artificial intelligence. For communications teams, complying with the regulations is one part of the job. The other is to maintain the quality of their relationship with the public.
This preparation begins before deciding which label to add. It requires an editorial policy that clarifies permitted uses, the level of review, transparency criteria and the person who will be held accountable if something goes wrong. This avoids two common extremes: concealing a relevant intervention or filling every piece of content with disclaimers that provide very little information.
Ultimately, an organisation demonstrates accountability when it can explain what role AI played, which decisions were retained by the human team, and how it will respond in the event of an error. Such clarity does not in itself guarantee trust, but it does enable a more honest relationship with the public.
If you work in institutional, public or social communication, I’d be interested to hear about your experience on LinkedIn: which decision regarding the use of AI remains unclear within your team?
It came into force on 1 August 2024. AI literacy and prohibited practices apply from 2 February 2025. The general application date is 2 August 2026, with exceptions.
Normally, the organisation is responsible when it uses an AI system under its authority in a professional capacity. The team is involved in that use, even if it did not develop the tool.
No. It depends on the type of content, the extent to which it has been generated or manipulated, its purpose, the potential for confusion and editorial review.
Article 50 provides for disclosure of its generation or manipulation by AI. An exception applies where such content has undergone human review or editorial oversight and a natural or legal person retains editorial responsibility.
Permitted and restricted uses, levels of review, transparency criteria, data protection, documentation of sensitive uses, and those responsible for approval and response.
This article is based on the topic, approach and editorial criteria defined by me. Artificial intelligence tools were used in its preparation to assist with research, organise the content and produce a first draft. Before publication, I reviewed the data, the linked sources, the context, the tone and the final wording. I, Laura Mellado, have approved this version and assume editorial responsibility for its content.
Media literacy should not sound like a lecture. Its value lies in helping people read, question, verify and participate more thoughtfully. Spotting a hoax is the starting point, not the destination.
Diverse voices in a multi-stakeholder programme are not the problem. The problem is the absence of an architecture that organises them without silencing them. How to design narrative coherence when multiple actors, logics and audiences must communicate the same programme without fragmenting or becoming uniform.
Institutions keep designing messages for public feeds while their audiences make decisions in WhatsApp groups no institutional algorithm can reach. The problem is not technical. It is a question about how trust works when public visibility disappears and what remains are relationships.