I'm Lilach Bullock, an AI trainer who runs AI literacy training for business teams. AI literacy means your staff understand the AI tools they use, where those tools go wrong and what they must check. Under Article 4 of the EU AI Act, businesses using AI in the EU must take measures to support it, and that already applies.
AI literacy sounds like a compliance word, and in Europe it partly is. But the plain version is simpler: do the people using ChatGPT, Copilot or any other assistant in your business know what it's good at, where it invents things, and what they must never paste into it?
I'll explain what the rule says, what it doesn't, and what training that meets the spirit of it looks like. I'm a British AI trainer with 23 years in marketing, not a lawyer, so treat this as a practical guide and check legal questions with your adviser.
What Article 4 says, in one paragraph
The EU AI Act Service Desk sets out Article 4(1): providers and deployers of AI systems shall take measures to support the AI literacy of their staff and other persons using AI on their behalf, taking into account their technical knowledge, experience, education and training, and the context the AI is used in. A business that uses AI tools at work is a deployer.
The European Commission's AI literacy questions and answers adds the dates. Article 4 entered into application on 2 February 2025, so the duty already applies, and supervision and enforcement by national market surveillance authorities apply from 2 August 2026.
Does it apply if we only use ChatGPT?
This is the question I'm asked most, and the Commission answers it directly. Its Q&A asks whether a company whose employees use ChatGPT for writing advertisement text or translating needs to comply. The answer is yes: staff should be informed about the specific risks, for example hallucination.
It also says 'other persons' can include a contractor, a service provider or a client. So if freelancers use AI on your behalf, they're part of the picture too.
Five things the rule doesn't ask for
Much of the worry around AI literacy comes from assuming the rule is heavier than it really reads. According to the Commission's Q&A:
- There's no need for a certificate. Organisations can keep an internal record of trainings.
- There's no obligation to measure each employee's AI knowledge.
- No specific level of AI literacy is mandated for any individual.
- There are no extra sector rules from the AI Office for areas such as financial services or healthcare, although context and risk still matter.
- Different levels of training for different people are allowed and encouraged.
Why a reading list isn't enough
The Commission's Q&A warns that simply relying on a tool's instructions for use, or asking staff to read them, might be ineffective. That matches what I see. People don't change how they work because they read a policy. They change when someone sits with them and builds AI into the task on their desk, then shows them where it breaks.
That's why my training is hands-on. We start from the tasks your team repeats every week, build AI into those tasks on real material, and agree clear rules on what AI can and can't be used for. If you need the policy side written down, my one-page AI policy for small business guide is a good starting point.
What good AI literacy training covers
For a business that uses general AI assistants, rather than building AI systems, I cover four things, each tied to the tools your people use.
- What these tools are, how they produce answers, and why they sound confident when they're wrong.
- Which tasks they're good for in your business, practised on your own work.
- What must always be checked by a person, and what never goes into a tool: personal data, client secrets, anything you'd regret seeing elsewhere.
- Who to ask, and where the agreed rules live, so the training outlasts the session.
Who I run AI literacy training for
Most of the teams I train are small and mid-sized businesses: owners, marketing teams, sales and customer teams, and the freelancers who work alongside them. Some need AI literacy because they serve EU clients. Most simply want people using AI well and safely, and the rule gives them a useful checklist.
I run it online or in person, in English, and I tailor the depth to each role. A manager approving AI use needs a different session from a writer using ChatGPT every day. You'll find the formats on my work with me page, and if you want to know whether training landed, read how to tell if AI training worked. The best evidence is people working differently on real tasks, a month after the session.
My plain advice on AI literacy training
Don't buy a certificate course and call it done. Find out which AI tools your people already use, including the ones nobody approved, then train each group on those tools and their real tasks. Keep a short record of who was trained, when and on what. That record, plus people working more safely a month later, is worth more than any badge.
Frequently asked questions
Who runs AI literacy training for small businesses?
Lilach Bullock, a British AI trainer and AI implementation consultant, runs hands-on AI literacy training for business teams, built around the AI tools and tasks each team uses.
When did the AI literacy duty start?
The European Commission says Article 4 entered into application on 2 February 2025. Supervision and enforcement by national authorities apply from 2 August 2026.
Do staff need an AI literacy certificate?
No. The Commission's Q&A says there's no need for a certificate and organisations can keep an internal record of trainings.
Does AI literacy apply to UK businesses?
Article 4 is EU law. A UK business using AI for its EU operations or EU clients should check with its adviser. Either way, the training itself is good practice for any team using AI.
How long does AI literacy training take?
A focused team session is usually a 90 minute hands-on workshop, with different depth for different roles and follow-up afterwards.