Orbrya
All posts
Orbrya2026-03-12

What EU AI Frameworks Teach Us About AI Literacy

Europe has been building formal AI literacy frameworks for years. What families can learn from DigComp and the AILit Framework about teaching children to evaluate AI.

When parents in the United States ask what AI literacy should actually look like for children, there's no national framework to point to. There's no federal curriculum, no standardized competencies, no government-produced guide for how families or schools should approach teaching children to work critically with AI. There are individual schools doing interesting things, and a growing body of research, but no coordinated answer.

Europe has been working on this problem longer and more systematically. The European Union's approach to digital and AI literacy, particularly the DigComp framework and the emerging AILit Framework, offers something American families can actually use: a structured way of thinking about what AI competence involves at different levels, and what skills are foundational versus advanced.

This post isn't a political argument for any particular policy approach. It's a practical look at what those frameworks describe and what families can take from them.

DigComp: A competence map for the digital world

DigComp, now in its third version (DigComp 3.0), was developed by the European Commission's Joint Research Centre as a reference framework for digital competence. It identifies five broad competence areas: information and data literacy, communication and collaboration, digital content creation, safety, and problem solving.

The information and data literacy area is most directly relevant to AI literacy, and it's more specific than its name suggests. It describes competencies including the ability to "browse, search, filter data, information, and digital content," "evaluate data, information, and digital content," and "manage data, information, and digital content."

The evaluation competency is where AI enters most acutely. DigComp describes it as analyzing, comparing, and critically evaluating the credibility and reliability of sources of data, information, and digital content. In the context of AI, this means not just asking "did AI give me an answer" but asking whether the answer is from a reliable underlying source, whether the confidence AI expressed is warranted, and whether the information serves the purpose it's being put to.

What DigComp captures well is that these skills are not AI-specific; they're digital literacy skills that AI makes newly urgent. A child who has learned to evaluate information sources in general is better positioned to evaluate AI output specifically. The habit of interrogating where a claim comes from transfers across contexts.

The AILit Framework: Competencies specific to AI

The AILit Framework goes further, focusing specifically on AI literacy as a distinct competency domain. Developed as a joint initiative of the European Commission and the OECD, with support from Code.org and international experts, a draft was released in May 2025 with the final version expected in 2026. It identifies several dimensions of AI literacy that go beyond general digital competence.

The dimension most applicable to families is what the framework calls "critically evaluating AI" -- the ability to assess AI outputs for accuracy, bias, and reliability, and to understand the limitations of AI systems. This maps directly to what Chahna Gonsalves described as critical thinking toward the AI rather than just for the assignment.

The framework also identifies "AI ethics and society" as a competency area, which covers understanding how AI systems can reflect the biases of their training data, how AI decisions affect different groups differently, and how to reason about the social implications of AI use. For older children, this is not abstract: it connects directly to real cases, including a 2023 peer-reviewed Stanford study finding that seven widely used AI detection tools falsely flagged over 61% of essays written by non-native English speakers as AI-generated, while performing near-perfectly on native speakers' essays. (Caveat: the study tested 91 essays from a single Chinese educational forum, which is a small and specific sample, and detection tools have been updated since April 2023.)

A third dimension -- "AI and human agency" -- describes understanding the appropriate roles of AI and human judgment in different contexts. When is AI a useful tool and when does relying on it compromise the quality of thinking or decision-making? That question is as relevant for a twelve-year-old doing homework as it is for a professional preparing a report.

What families can take from these frameworks

Neither DigComp nor AILit was written for home educators, and neither translates directly into a lesson plan. But they offer something useful: a vocabulary and structure for thinking about AI literacy as a set of learnable competencies rather than a vague aspiration.

The frameworks suggest that AI literacy is not primarily about knowing how AI works technically. It's about being able to evaluate AI output, understand AI limitations, and make informed decisions about when and how to use AI and when not to. Those are judgment skills, not technical ones, and they develop through practice over time.

They also suggest that AI literacy is developmental -- that the competencies expected of a foundation-level learner are genuinely different from those expected at an advanced level, not just in sophistication but in kind. A child learning to recognize that AI can be wrong is at a different stage than a teenager learning to analyze how AI training data shapes output in ways that reflect particular perspectives or reinforce particular biases.

For a family building AI literacy at home, the practical application is a simple question to revisit periodically: are we building habits that help our child evaluate AI outputs, or are we only teaching them to use AI more effectively? Both matter. But the former is the one that schools are not yet systematically providing, and the former is the one that will remain valuable as AI tools continue to change in ways that make any specific tool-use knowledge obsolete.

The European frameworks exist because their authors recognized that AI literacy is not something that happens incidentally. It has to be taught, practiced, and built up over time. That recognition is exactly right -- and it's available to families without waiting for a national curriculum to arrive.

Sources cited in this post: