The most common EU AI Act mistakes in marketing and PR are assuming the law only applies to European companies, believing it only concerns AI developers, labelling nothing or labelling everything, keeping no records of AI decisions and treating staff training as optional. Each mistake is easy to make and fairly easy to fix once a team knows what the regulation actually asks.
The EU AI Act is the European Union’s regulation on artificial intelligence, adopted in 2024, with obligations that phase in over several years. It sorts AI uses by risk, bans a short list of practices, sets heavy rules for high-risk systems and adds transparency duties for chatbots and synthetic media. Most marketing work sits in the transparency layer, which is where most mistakes happen.
Mistake 1: does the Act only apply to EU companies?
No. The Act applies to providers placing AI systems on the EU market and to deployers using AI systems in the EU. It also reaches providers and deployers outside the EU when the output of their AI system is used in the Union.
In practice, a Turkish agency producing AI-generated video for a French client’s campaign, or running a chatbot on a website aimed at German customers, should assume it is in scope. Clients in the EU increasingly add AI clauses to agency contracts for this reason.
Mistake 2: “we only use AI, we don’t build it”
Many teams think obligations fall only on companies like OpenAI or Google. The Act distinguishes providers, who develop or place systems on the market, from deployers, who use them under their own authority. Deployers have their own duties.
For marketing teams as deployers, the main duties are telling people when they interact with an AI system where this is not obvious, disclosing deepfakes, ensuring staff AI literacy and, for any high-risk use, following the provider’s instructions and keeping human oversight. A company can also become a provider by putting its own name on a system or changing its intended purpose substantially.
Mistake 3: why is labelling everything a problem too?
Some teams react to the Act by stamping “AI-generated” on every asset touched by AI, including a headline polished by a writing assistant. That dilutes the signal and can make audiences distrust ordinary content.
The disclosure duty targets specific cases: AI systems that interact with people, and image, audio or video content that resembles real people, objects, places or events and could falsely appear authentic. Text published to inform the public on matters of public interest also carries a disclosure duty when AI-generated, unless it has gone through human review and editorial responsibility.
A practical labelling rule
- Chatbots and voice assistants: always disclose at the start of the interaction
- Realistic synthetic images, video and voices of people or events: always disclose
- Clearly artistic, satirical or fictional work: disclose in a way that does not spoil the work
- AI-assisted copy edited and approved by a person: no label needed under the transparency rule, but follow client policy
Mistake 4: no record of decisions
When a client’s legal team asks why a tool was approved, many agencies have only an email thread. That is weak evidence and hard to find later.
Keep a one-page record per AI system in production: what it does, the vendor, the risk classification and reasoning, disclosure measures, the approver and the date. Review it when the tool or its use changes. This also helps with GDPR or KVKK questions, since many AI tools process personal data.
Mistake 5: skipping AI literacy training
The Act requires providers and deployers to take measures to ensure a sufficient level of AI literacy among staff dealing with AI systems. This obligation applies early in the phase-in schedule, from February 2025.
Training does not need to be long. A one-hour session per role, covering what the tools can and cannot do, where they make mistakes, what data must not go in and how disclosure works, meets the spirit of the rule better than a generic online course. Keep attendance records, and repeat the session when a major new tool is introduced.
Mistake 6: missing high-risk uses hidden in marketing
Most marketing AI is not high-risk, which leads teams to assume none of it is. Some uses cross the line. AI that screens job applicants for an employer branding campaign, scores people for credit or insurance offers, or decides access to essential services can fall into high-risk categories listed in the Act.
Emotion recognition and biometric categorisation also deserve care. Some uses are prohibited, such as emotion recognition in workplaces and schools, with narrow exceptions. A “smart billboard” that reads faces to infer mood should go to legal review before anyone builds it.
- Recruitment and HR screening tools
- Credit, insurance and pricing decisions about individuals
- Biometric identification or categorisation
- Emotion recognition in workplaces or education
How Alien Road avoids these mistakes in client work
In a typical engagement, discovery includes a short inventory of every AI tool used across content, social, paid media and customer service, with a flag for EU audiences or EU clients. We note any use near a high-risk category and send it to the client’s counsel.
Setup covers disclosure wording for chatbots and synthetic media, a labelling rule written into brand guidelines, one-page decision records and role-based training sessions with attendance logs. Reporting is quarterly: new tools added, records updated and any change in how a tool is used. Legal interpretation always stays with qualified counsel; our job is making the marketing side documented and consistent. When a client operates in several EU countries, we also check whether local advertising codes add their own disclosure expectations on top of the Act.
Where to start this week
Ask each team lead one question: which AI tools did you use last month, and does any output reach people in the EU? Collect the answers in one sheet.
Then write two disclosure lines, one for chatbots and one for realistic synthetic media, and get them approved. Those two lines close the most visible gaps, and the rest of the work can follow at a steady pace as the obligations phase in.
Keep reading: PR Services · Digital PR