AI is eroding trust in information. Not in some abstract, philosophical sense. In a measurable, operational sense that affects workplaces, governments, and classrooms right now.
Management consultancies including McKinsey, BCG, and Deloitte have all published research showing the same thing: generative AI is blurring authorship and destroying confidence in the provenance, accuracy, and reliability of written content. Workers and managers increasingly cannot tell whether the document in front of them was written by a human, generated by AI, or some combination of the two. And they cannot tell whether it is accurate.
The Council of Europe warned in March 2026 that if education systems fail to build strong critical reasoning skills, organisations will develop what they called "systemic cognitive vulnerabilities." The inability to distinguish between genuine human insight and machine-generated hallucination. Not as a theoretical risk. As an operational reality.
This is already happening in schools
A teacher sets an essay. Half the class submits work that reads fluently but says nothing specific. The sentences are grammatically perfect. The arguments are structurally sound. The evidence is plausible but unverifiable. The teacher suspects AI was involved but cannot prove it, because the writing is good enough to pass and generic enough to be undetectable.
That scenario is playing out in thousands of classrooms every week. And the response in most schools is to focus on detection. Can we catch them? Can we prove it?
That is the wrong question. The right question is: do our students understand what they lose when they let AI do their thinking?
The authorship problem runs deeper than cheating
When a student submits AI-generated work, the obvious concern is academic integrity. But the deeper problem is epistemological. The student does not know what they think. They have not done the cognitive work of forming an argument, testing it against evidence, and defending it. They have a finished product with no understanding underneath it.
Scale that across an entire generation and you get exactly what the consultancies are warning about. A workforce that produces polished outputs without understanding whether those outputs are true, accurate, or meaningful.
The African Union's intelligence briefings from March 2026 put a sharper edge on it. AI-assisted biometric fraud surged 87% across Southern Africa. Deepfakes. Digital impersonation. Identity theft. When people cannot distinguish real from generated, the consequences are not academic. They are financial, legal, and in some cases physical.
What schools should be teaching instead
The Council of Europe's framework for AI literacy puts the "human dimension" first. Before the technology. Before the practical skills. The human dimension covers ethics, critical thinking, cultural awareness, democratic participation, and environmental impact.
Their argument is simple. If you teach students to use AI without teaching them to question it, you have not taught literacy. You have taught compliance.
UNESCO made the same point in April when they launched their Observatory on AI in Education for Latin America. Their founding principle: "Education must govern AI, not the other way around." They warned explicitly that uncritical adoption of AI tools risks deprofessionalising teachers and eroding pedagogical autonomy.
Both organisations are saying the same thing. The critical thinking has to come first. The tool skills come second.
Every subject has a role
This is not a Computing problem and it is not a PSHE problem. It is an every-subject problem.
A History teacher who teaches source evaluation is teaching students to question provenance and bias. Apply that to AI-generated content and you have an AI literacy lesson. An English teacher who teaches persuasive writing techniques is one step away from teaching students how AI generates persuasive text and why that matters. A Science teacher who teaches experimental method is building the exact skill set students need to question whether an AI-generated claim is supported by evidence.
The critical reasoning skills already exist in the curriculum. They just need to be connected to AI.
AILitKit makes that connection. You put in your lesson. The tool identifies where critical AI literacy lives inside it and gives you the activities to make it explicit. Coaching notes so you know what to say. Support, challenge and differentiation so every student can access it. Framework alignment so your school can evidence delivery.
Because the question is not whether your students will encounter AI-generated content. They already are. Every day. The question is whether they have the skills to look at it critically, or whether they accept it at face value because nobody ever taught them not to.
Matthew Wemyss is the founder of AILitKit and IN&ED, and author of AI in Education: An Educator's Handbook.