In this article
The short answer
A practical AI literacy framework for employees has eight competencies: understanding how AI works, knowing where AI is in your tools, asking for useful work, checking output, protecting data, recognising bias, following policy and accountability, and knowing when not to use AI. Each is assessed at four levels, from aware to champion, and each role is assigned a target level.
Frameworks have a bad reputation because many are forty pages long and nobody reads them. This one fits on a page. Use it to agree what "AI literate" means in your business, then train and assess against it.
What are the eight AI literacy competencies?
1. Understanding how AI works
Explains in plain English that generative AI predicts likely text or images from patterns, can be confidently wrong and does not know your business unless told. Background: what is generative AI.
What good looks like
Explains to a colleague why AI output must be checked, without jargon.
2. Knowing where AI is in your tools
Recognises AI features in everyday software, such as Copilot in Microsoft 365, meeting note-takers and AI inside sector systems, and knows which are approved.
What good looks like
Can list the AI features in their own daily software and which are approved.
3. Asking for useful work
Writes clear requests with role, task, context, format and limits, supplies the real material and iterates. See how to write prompts for AI.
What good looks like
Gets a usable first draft within two or three turns on a routine task.
4. Checking output
Verifies facts, figures, names, quotes and sources; spots missing information and weak reasoning; matches the effort of checking to the stakes.
What good looks like
Catches an invented figure or quote before it leaves the building.
5. Protecting data and confidentiality
Knows what must never go into which tool, removes identifiers, understands the difference between consumer and business accounts. See how to keep information safe using AI tools.
What good looks like
Removes names and identifiers by habit and uses only business accounts for work.
6. Recognising bias and fairness risks
Understands that AI can reflect and amplify bias, especially in decisions about people, and knows when to escalate. See using AI ethically in business.
What good looks like
Questions AI suggestions that affect people and escalates when unsure.
7. Policy, accountability and disclosure
Follows the company AI policy, knows that the person using AI owns the result, and discloses AI use where the policy or the client expects it.
What good looks like
Can explain the policy in their own words and says when AI helped.
8. Knowing when not to use AI
Recognises tasks where AI adds risk or no value: final decisions about people, sensitive conversations, work where checking takes longer than doing it yourself.
What good looks like
Chooses not to use AI for a sensitive conversation or a final decision about a person.
What are the four levels of AI literacy?
| Level | Description | What you would observe |
|---|---|---|
| 1. Aware | Knows the basics and the rules | Can explain what AI is, names approved tools, knows what not to paste |
| 2. Capable | Uses AI safely for routine tasks | Gets useful drafts, checks key facts, follows the policy without prompting |
| 3. Confident | Uses AI well across their role | Builds reusable prompts, uses documents as context, spots subtle errors and bias |
| 4. Champion | Helps others and improves practice | Coaches colleagues, maintains the team prompt library, flags policy gaps |
What level of AI literacy does each role need?
Set a target level per competency for each role. Here is a typical starting point; adjust for your own risk.
| Role | Competencies 1 to 4 (use) | Competencies 5 to 8 (protect and judge) |
|---|---|---|
| All staff | Aware | Capable |
| Regular AI users (marketing, admin, sales, ops) | Capable to Confident | Capable |
| Managers and approvers | Capable | Confident |
| HR, recruitment, finance, legal, customer decisions | Capable | Confident to Champion |
| AI champions | Champion | Confident |
| Directors | Aware to Capable | Confident |
Notice that the protect-and-judge competencies are often more important than the use competencies. A brilliant prompter who pastes client data into the wrong tool is a liability.
How do you use an AI literacy framework in practice?
Adapt it
Rename or merge competencies to fit your language. Add sector rules, such as SRA, FCA or safeguarding requirements.Baseline against it
Score each person on the eight competencies. The AI skills matrix template uses a compatible 0 to 3 scale.Train to the gaps
Design tiers around the largest gaps, as in our 90-day programme.Assess and record
Use scenarios and work samples, not just attendance. See how to measure AI literacy.Review yearly
Tools change quickly. Update the examples, not necessarily the competencies.
How does this compare with other AI frameworks?
Formal frameworks such as the NIST AI Risk Management Framework and ISO/IEC 42001 deal with organisational AI governance. They expect people to be competent but do not tell you what a receptionist needs to know. A workforce literacy framework like this one fills that gap and sits underneath your AI governance framework.
AI literacy frameworks: questions people ask
What should an AI literacy framework include?
Clear competencies, such as understanding AI, using it, checking output, protecting data, recognising bias, following policy and knowing when not to use AI, plus proficiency levels and target levels for each role.
How many levels should an AI literacy framework have?
Three or four is usually enough. This framework uses four: aware, capable, confident and champion.
Is there an official AI literacy framework in the UK?
There is no single mandatory UK framework for workplace AI literacy. Organisations generally adapt published guidance and frameworks to their own roles and risks.
Can we use this AI literacy framework in our company?
Yes. It is written to be adapted. Rename the competencies to fit your language, add any sector-specific rules and set target levels for your roles.
How is an AI literacy framework different from an AI skills matrix?
The framework defines what good looks like. The skills matrix records where each person currently is against it, so you can plan training.
Keep reading
Training built on a framework your team can see
We map your roles to the eight competencies, train to the gaps and leave you with the framework, the prompt library and the records.
- Basic AI Training Hands-on AI literacy for every employee.
- AI Policy Workshop Write the rules your training will teach.
- AI Capability Assessment Baseline your team before you train.
Harry Lang
Harry Lang founded The Oxford AI School in Witney, Oxfordshire. He trains UK businesses and teams to use AI well, writes the Marketing Unfiltered newsletter, and is the author of Learn To Fly With AI.