AI literacy in business: what has mattered since February 2025

Artificial intelligence is already used in many businesses, even where nobody has agreed an official AI strategy. Since the beginning of February 2025, the EU AI Act has added a new dimension to the subject. Does every employee now need an AI driving licence? Or could a few clear everyday rules be enough? This article looks at what AI literacy means and what a pragmatic approach might involve.

Why AI literacy has become a management issue

Artificial intelligence often enters a business faster than an official strategy. One person has an email drafted, another summarises meeting notes, and the marketing team generates its first image ideas. This can be useful. Difficulties arise when nobody specifies what information may be entered into the tools, who checks the results or who remains responsible for mistakes.

Parts of the EU AI Act, including provisions on AI literacy, have applied since 2 February 2025. These requirements do not apply indiscriminately to every private use. For providers and deployers of AI systems, the task is to take measures, to the best of their abilities and in light of the context of use, to ensure appropriate knowledge among the people involved. This does not create a universal state-mandated certificate for every employee.

What employees actually need to know

Training does not have to begin with complicated mathematics. Four foundations help in everyday work: understanding what an AI system can do, recognising typical errors, protecting confidential information and checking outputs before publication. Anyone receiving a suggested text should understand that plausible wording does not make the facts reliable.

Customer service also involves data protection and polite communication. Marketing raises questions about brand voice, copyright, image rights and advertising claims. Someone integrating AI into a process through an API needs different knowledge from someone using a wording assistant. The measures should reflect these different tasks.

What a simple AI policy could look like

Start with a one-page document. List approved tools, prohibited data types, responsible people and a clear rule for human review. Specify that personal customer data and internal calculations must not be copied into external chat services without assessment. Add examples of tasks where generated content requires a subject-matter expert's approval before use.

A practical starting point is a short workshop built around three real tasks: a product description, a summary and an image draft. Participants can examine together where AI helps and what checks remain necessary. Record their questions and update the rules when tools or uses change.

Practical example: AI in a small marketing team

A four-person team wants to use AI to brainstorm social-media ideas. Management selects an approved tool, a folder containing non-confidential product information and an approval process. The AI may propose ten ideas; publication happens only after a responsible person has checked claims, tone and image rights. Once a quarter, the team discusses problematic outputs.

The workload remains manageable, but the value is greater than an isolated AI training session with no connection to daily work. Businesses should involve specialists for legally sensitive uses and assess the relevant rules separately.

Conclusion

AI literacy is more than a box on a checklist. What matters is that people understand their tools and intervene at the right moment. My suggestion is to begin with your team's actual tasks, document the main rules and create an easy way to report questions or mistakes. A new compliance topic can then become a useful part of how the business works.

Sources and further reading

The following original sources and official documentation provide further background. Always check the current position for legal and other changing topics.

• eur-lex.europa.eu

Note: This article provides general information in a German or European context and does not replace individual legal advice.