AI literacy for freelancers: why this applies to you too

As a freelancer you read about the AI Act and quickly think: that is about companies with staff, not about me. Understandable — much of the attention goes to employers who need to train their employees. But if you work for clients, you are not outside this story. In fact, as a one-person business, you are your entire organisation. Every AI decision you make, you make on behalf of your business — and it directly affects the work you deliver to clients.

Why Article 4 thinking applies to you too

Article 4 of the AI Act asks organisations that deploy AI to ensure the people working with it have a sufficient level of AI literacy. The provision was written with “providers and deployers” in mind — and a one-person business using AI professionally does not automatically fall outside that. The precise legal scope for the very smallest businesses is debatable, and we won’t make a stronger claim here than is responsible. But there are three practical reasons why it is wise to think as if it applies to you:

• Your clients do fall under it. If you work for an organisation with staff, that organisation has a literacy obligation for everyone working with AI on its behalf. More and more clients therefore set requirements for externals too: how do you handle AI in the work you do for us?
• You carry the full risk yourself. When poorly checked AI output causes a mistake, there is no employer to absorb the blow. A wrong figure in a report, an invented source in an advisory document, leaked client information — it all lands on your name and your liability.
• The norm is shifting. Even apart from the letter of the law, “I didn’t know AI could get that wrong” is becoming an ever weaker defence. The AI Act makes AI literacy the professional standard, and clients will hold suppliers to that standard too.

How the AI Act phases in over time is covered in our overview of the AI Act timeline.

Client data: your responsibility, not the tool’s

The biggest concrete risk for freelancers is not fines, but data. If you paste a client’s draft contract into a public AI tool to have it summarised, you are sharing confidential information with an external party. Depending on the tool and its settings, that information may be stored or used for training. If it contains personal data, the GDPR comes into play — and in many cases you also have a confidentiality agreement with your client that you are breaching at that moment.

A few rules of thumb you can adopt today:

• No client data in public AI tools, unless you have checked what happens with input and your client agrees. Anonymising (stripping names, amounts and identifying details) is often a workable alternative.
• Check your tools’ settings. Many AI services have an option to prevent your input being used for training. Turn it on, and record that you did.
• Be transparent with your client. Simply explain how you use AI and what you don’t use it for. It prevents trouble afterwards and — let’s be honest — it is also just a good sales conversation: you show that you handle this more deliberately than the average competitor.
• Check everything that goes out the door. AI output can be convincing and wrong at the same time. What AI structurally cannot do, and why knowing that is precisely your craft, is covered in our article on the limits of AI.

Stand out with demonstrable skill

Here is the opportunity. Almost every freelancer now puts “works with AI” on their website. So that no longer says anything. What does say something: being able to explain how you use AI, where your boundaries are and how you safeguard quality — and being able to back that up with something verifiable.

Consider the difference between these two sentences in a proposal:

• “I use AI to work more efficiently.”
• “I use AI for draft versions and research, check all output manually, never enter confidential client data into public tools, and have made my AI literacy demonstrable with a certificate you can verify online.”

The second sentence wins the conversation. Not because the certificate is magic, but because you show that you have thought about the risks — exactly what clients, especially larger ones, are nervous about right now. To them, an external who has their AI use in order is one less thing to worry about.

Keep your own evidence

Organisations record their AI literacy efforts in a file. As a freelancer, the one-person version of that is sensible and easy to keep small:

1. What you have learned: which course or training, when completed, with the certificate or verification link attached.
2. How you work: a short AI working method of your own on one page — which tools you use, for what, which data does and does not go in, how you check output. You can show this document to a client the moment they ask.
3. What you have agreed: if a client sets specific requirements for AI use in an assignment, record them in the engagement confirmation.

This costs you an afternoon, and it is exactly the kind of evidence that makes the difference when a client, insurer or dispute asks for it. Writing down your working method also forces you to really think it through — most freelancers discover at least one habit they want to change in the process.

Where to start?

If you want to know where you stand right now, start with the free quiz — in a few minutes it gives you a picture of your knowledge about how AI works, where it goes wrong and what the rules are. Score high? Great — your one-page working method will practically write itself. Score lower than you thought? Then you know exactly which knowledge you are missing, before a client finds out.

And if you work a lot for organisations with staff, feel free to point your clients to their own obligations — they can read more on our employers page. It makes you a conversation partner instead of just a supplier.