On November 5, 2026, the European Commission presented an Education and Skills Package built around four priorities: basic skills, digital and AI tools for education, the teaching profession, and vocational training. One line on the Commission’s package page concerns classrooms directly. The Commission plans to pilot an AI Literacy Certificate for 15-year-old learners in the 2027/2028 school year, alongside work to help pupils and teachers use AI safely and responsibly and to assess AI literacy skills.
The same page lists a figure under “Why this matters”: according to Eurostat, 64% of 16- to 24-year-olds use generative AI. Use is already widespread while the Commission is still planning how to assess it.
A certificate records what a pupil can do. The decision to place an AI tool in that pupil’s classroom, and the authority to withdraw it when it stops helping, rest with an adult the certificate never names.
What the Commission announced
The digital strand covers a new infrastructure consortium, the AI literacy work, and a Healthy Screens Pledge for schools and families. On November 3, the Commission set up the European Digital Infrastructure Consortium on Education, known as EDU-EDIC, at the request of eight founding Member States, to pool resources for trusted AI and digital services for schools and secure data exchange between institutions. Speaking about the package as a whole, Commission President Ursula von der Leyen said the Commission is backing Member States while “equipping children with the skills they need to safely navigate AI.”
On November 3 as well, Cedefop described a joint report from six bodies, including the European Commission, the International Labour Organization, and UNESCO, whose central message is that AI literacy is becoming a basic competence. Cedefop splits it into technical, analytical, and ethical literacy, and says that together they help people judge whether AI outputs are reliable, recognise risks, and “decide when human oversight is needed.” Its own 2024 survey across 11 EU countries found 40 to 60% of adult workers had low AI literacy, depending on the competence assessed, and only around 15% had taken AI-related training in 2023 and 2024.
Read together, the two announcements put the same skill at the center. For adults it’s knowing when a human has to step in. For pupils, the Commission’s answer is a certificate.
Where the framework puts the checkpoint
The Commission and the OECD published the AI Literacy Framework for primary and secondary schools earlier this year, and it contributes to the PISA 2029 Media and Artificial Intelligence Literacy assessment. One of its four domains, Manage AI, is described as “divide work intentionally between humans and AI.” The framework says managing AI “starts with students deciding if AI is truly needed.” At the advanced level of one Manage AI competence, learners “establish checkpoints based on human and AI roles, monitor progress against success criteria and adjust roles accordingly.” The framework also states that it is non-binding.
That is the right skill to teach, and it’s also where the checkpoint stays: inside the learner. In June, our review of the framework, The Continued Failure in AI Literacy, argued that AILit houses the checkpoint inside the learner’s own self-monitoring, so it works as a study habit with no named institutional human behind it. The package page doesn’t yet say how the certificate will be assessed. If it draws on that domain, it can confirm the habit in a 15-year-old, but it can’t supply the institutional half.
The HEQ paper, Measuring the Governance Competence at the Center of AI Literacy, defines the competence a pupil would be building as the ability to direct the work, challenge it, verify it, and own the result when AI is part of it. It defines the checkpoint itself as a hard stop where a human holds authority and decides to accept, modify, or reject. A pupil who stops using a chatbot on an essay has made a good individual call. A school that keeps a tool on every pupil’s device has made an institutional call, and someone should be able to say who made it.
This is also where Responsible AI and AI Governance part ways. Whatever age filters or data protections a trusted school AI service includes are Responsible AI: machine checking machine. They matter, and they run without a person deciding each case. AI Governance begins when a named human at the school or authority accepts, modifies, or rejects the use of that service for a group of pupils and a set of tasks, and the decision is written down.
What the pilot can measure, and what the school has to own
The pilot is planned for the 2027/2028 school year, which still leaves time to design the assessment and the accountability together. Three steps would help.
First, score the decision to use AI. A certificate task can ask a pupil to choose whether AI should do all of the work, part of it, or none of it, and to give a reason in writing. If the certificate follows the Manage AI domain, that task tests it directly, and it rewards a pupil for not using a tool when the task does not need one.
Second, name the signer at the school. Each school or authority taking part in the pilot should record one person who signs off on which AI services pupils use and for which tasks, with stated conditions for pausing or withdrawing a service. Each review gets an accept, modify, or reject entry with a date and a reason. The certificate would assess the pupil, while the decision record would show who approved the AI services in the room where the pupil learned and was tested.
Third, track a small set of indicators from the first pilot year: the share of participating schools with a named signer on file, the share of certificate tasks where a pupil correctly chose not to use AI and explained why, and the number of service decisions modified or rejected each term. A school that never modifies or rejects anything should ask whether its approvals involve any review at all.
The Commission plans to pilot the certificate in 2027/2028. The test of that year will be whether each participating school can put a name next to the decision to let AI into the room.
Sources
- Cedefop. (2026, November 3). AI literacy: The new basic skill for Europe’s workforce. https://www.cedefop.europa.eu/en/news/ai-literacy-new-basic-skill-europes-workforce
- European Commission. (2026, November 5). Commission launches Education and Skills package to tackle declining basic skills, rapid digital change and teacher shortages [Press release]. https://digital-strategy.ec.europa.eu/en/news/commission-launches-education-and-skills-package-tackle-declining-basic-skills-rapid-digital-change
- European Commission. (2026). Education and Skills Package. European Education Area. https://education.ec.europa.eu/focus-topics/improving-quality/education-and-skills-package
- European Commission & OECD. (2026). Empowering learners for the age of AI: An AI literacy framework for primary and secondary education. https://ailiteracyframework.org/pdfs/framework_pdf/AILF_en.pdf
- Puglisi, B. C. (2026, June 21). The continued failure in AI literacy: AILit produced a starting point halfway through the race and called theory a framework. https://basilpuglisi.com/ailit-framework-rule-optional/
- Puglisi, B. C. (2026, September 13). Measuring the governance competence at the center of AI literacy. https://basilpuglisi.com/measuring-governance-of-ai-literacy/
FAQ
What is the EU AI Literacy Certificate?
It is a proposed certificate the European Commission plans to pilot for 15-year-old learners in the 2027/2028 school year, announced on November 5, 2026 as part of the Education and Skills Package. The package also includes work to help pupils and teachers use AI safely and responsibly and to assess AI literacy skills.
How does the certificate relate to the EU and OECD AI Literacy Framework?
The Commission’s package page does not yet set out the assessment design. The closest published reference is the AI Literacy Framework the Commission and the OECD released for primary and secondary schools in 2026, which organizes competences into four domains (Engage with AI, Create with AI, Manage AI, and Shape AI) and states that it is non-binding.
What does the Manage AI domain ask pupils to do?
It asks learners to divide work intentionally between humans and AI, starting with whether AI is truly needed. At the advanced level of one competence, learners establish checkpoints based on human and AI roles, monitor progress against success criteria, and adjust roles accordingly.
Why is a pupil’s checkpoint not enough on its own?
A checkpoint inside a learner’s own self-monitoring is a personal habit. The decision to put an AI service in front of a class is an institutional one, and AI Governance needs a named human who accepts, modifies, or rejects that use and records the decision.
How is Responsible AI different from AI Governance in a school?
Responsible AI means machine checking machine, for example filters and data protections built into an education service. AI Governance adds a named person at the school or authority who decides whether the service is used, for which tasks, and when it is paused or withdrawn.
What should schools in the pilot measure?
Useful indicators include the share of participating schools with a named signer on file, the share of certificate tasks where pupils correctly chose not to use AI and explained why, and how many AI service decisions were modified or rejected each term.
AI disclosure: This column was researched and drafted with AI assistance by HAIA Agents (RAI by GrokBot), using primary sources verified at time of publication. #AIgenerated
Leave a Reply