The EU AI Act affects PR and communications teams mainly through transparency duties: labelling deepfakes, disclosing certain AI-generated text, and knowing which AI tools the team uses and why. Practical workflows for an inventory, a disclosure decision, a review log and vendor checks save time because the same questions stop being debated on every campaign.
The AI Act is the European Union’s regulation on artificial intelligence. It entered into force in 2024 and applies in phases, with bans on certain practices first, rules for general-purpose AI models next, and transparency and high-risk obligations later. Dates and guidance have been subject to discussion and possible adjustment, so check the current timetable with legal counsel rather than relying on a blog post. This article covers working processes, not legal advice.
Why should a PR team care?
Most PR work falls outside the Act’s high-risk categories. But communications teams generate synthetic images, voices and video, publish text on matters of public interest and run campaigns into EU markets even when based outside the EU. The transparency rules in Article 50 speak directly to that work, including disclosure of deepfakes and of AI-generated text published to inform the public, with an exemption where content has gone through human review and someone holds editorial responsibility.
There is also a reputational side. Journalists and audiences increasingly ask whether an image or quote was AI-generated. A clear internal process lets the team answer quickly and honestly.
Workflow 1: build an AI tool inventory
You cannot apply rules to tools you do not know about. Start with a list.
- Ask every team member which AI tools they use, including features inside other software such as design apps and email tools.
- Record the tool, provider, what it is used for, what data goes into it and who approved it.
- Mark tools that create images, audio or video of real people.
- Note the provider’s own transparency features, for example watermarking or content credentials.
- Review the list every quarter; new features appear inside tools you already use.
Workflow 2: the disclosure decision
Instead of debating each asset, use a short decision path. Is the content an image, audio or video that resembles real people, places or events and could be mistaken for genuine? Then label it as artificially generated or manipulated. Is it text published to inform the public on matters of public interest? Then disclose AI generation unless a human reviewed it and the organisation takes editorial responsibility.
Record the decision with the asset. A spreadsheet column or a field in your digital asset manager is enough. When a journalist asks later, you can show what was decided and why.
Workflow 3: deepfake and synthetic media rules
PR teams face the highest risk with synthetic people. A generated “customer testimonial” video, a cloned executive voice, or a realistic image of an event that did not happen can cross from creative to deceptive.
- Never generate a realistic likeness of a real person without their written consent.
- Label synthetic media clearly in the asset itself and in captions, not only in metadata.
- Keep content credentials such as C2PA data intact when exporting, where tools support it.
- Prefer obviously illustrative styles when the goal is explanation rather than realism.
Workflow 4: the human review log
Because human review and editorial responsibility matter for AI-generated text, document them. For each press release, statement or article drafted with AI help, record who reviewed it, what they checked (facts, quotes, figures, claims about third parties) and who approved publication.
This log also protects quality. AI drafts sometimes invent quotes, misstate dates or attribute statements wrongly. A named reviewer who checks every factual claim against a source catches most of these before they become corrections.
Workflow 5: vendor and agency checks
If agencies, freelancers or platforms produce content for you, the obligations do not disappear. Add three questions to briefs and contracts: which AI tools will be used, how synthetic media will be labelled, and who reviews factual accuracy. Ask general-purpose AI providers for the documentation the Act expects them to make available to downstream users, where relevant to your use.
What are common mistakes?
- Treating a metadata watermark as sufficient disclosure when the audience never sees it.
- Labelling everything as AI-generated, which makes the label meaningless.
- Leaving AI use decisions to individuals with no shared rule.
- Assuming the Act does not apply because the company is outside the EU, while campaigns target EU audiences.
How do these workflows save time?
The first campaign under a new process feels slower. After that, the team stops arguing about each asset because the decision path answers the question in a minute. Reviewers know what to check, legal sees fewer last-minute requests, and the inventory makes it quick to respond when a tool changes its terms or a client asks how their content was made. Consistency is the time saving: the same rule applied every time is faster than a fresh debate every time.
How Alien Road applies EU AI Act workflows in PR work
In a typical PR engagement, discovery starts with the inventory: we list the AI tools used across the client’s team and ours, the content types they produce and the markets the content reaches. That shows quickly where transparency rules are likely to apply.
Setup turns the workflows above into templates: a disclosure decision checklist attached to every content brief, a labelling standard for synthetic visuals and audio, and a review log for AI-assisted text. Legal review of the templates stays with the client’s counsel.
Measurement is simple and auditable. Monthly reports show how many assets used AI, how many were labelled, review log completeness and any journalist or audience questions received. The goal is a process the client could show a regulator or reporter without preparation.
Where to start this week
Send a three-question survey to your communications team: which AI tools do you use, for what, and with what data. Combine the answers into an inventory spreadsheet. Then add a single “AI disclosure” field to your content approval form. Those two steps give you the base for every other workflow.
Keep reading: PR Services · Digital PR