The second winner of UNESCO's 2026 ICT in Education Prize wasn't a tech company. It was a Brazilian state government's education department. Redação Paraná — "redação" means essay or composition in Portuguese — is an AI writing-feedback programme run by the State Secretariat of Education of Paraná across its public schools. It's a useful case study for any parent trying to work out what a responsible, large-scale use of AI in a subject as personal as writing actually looks like.

Writing is one of the places parents worry most about AI — it's easy to imagine a tool that just writes the essay for a child rather than helping them write it themselves. A state government deploying AI writing feedback across its entire public-school system, and winning an international prize for doing it well, is worth a closer look.

What we actually know

UNESCO's announcement confirms two things clearly: Redação Paraná is run by the State Secretariat of Education of Paraná (a Brazilian state government body, not a private company), and it uses AI for writing feedback across the state's public schools. Beyond that, we don't have granular detail — no participant numbers, no description of the exact software, no evaluation data — and we won't invent any.

What we can reasonably infer, because it follows from what a government education department running this kind of programme typically has to do, is this: at public-school scale, a single teacher cannot give detailed feedback on every student's essay every week — the volume simply doesn't allow it. A programme built to solve that problem almost certainly uses AI to produce a fast first pass of feedback (structure, clarity, argument, mechanics), while the teacher remains the one who decides what the feedback actually means for that student's mark and next steps. Read that as an informed inference about how programmes like this typically work, not a confirmed description of Redação Paraná's internals.

Why "AI gives feedback, teacher decides" is the model worth noticing

This is the crux of what makes Redação Paraná worth studying rather than just filing away as a foreign news item. There's a meaningful difference between three things:

  1. AI writes the essay. This is the outcome every parent fears — and rightly. It removes the thinking entirely.
  2. AI grades the essay, final answer. Faster than a teacher, but it hands a high-stakes judgement to a system with no accountability for getting it wrong.
  3. AI gives a fast first pass, teacher decides. This is the model a public, accountable education system is most likely to have chosen — it uses AI for what AI is actually good at (speed, consistency, catching mechanical issues) while keeping a human responsible for the judgement that actually matters to a student's grade and confidence.

That third model is the one worth holding up against any AI writing or tutoring tool you're considering for your own child. Does it produce work for them, does it make a final unaccountable judgement about them, or does it help sharpen their own effort while a real person stays in charge of what matters?

The academic integrity question this raises

A programme built around AI feedback on student writing sits close to a question every UK parent of a secondary-age child is now grappling with: where's the line between AI helping with an essay and AI doing it? We've written about this directly in our piece on AI and academic integrity — where UNESCO draws the line. The short version: feedback that helps a student see what to fix themselves is a fundamentally different thing from a tool that fixes it for them.

What this means for you

You won't be able to sign your UK child up for Redação Paraná — it's a Brazilian state programme. But the model it represents is directly useful when you're evaluating any AI tool that touches your own child's writing, whether that's an English tutor, an essay-feedback app, or a general-purpose chatbot they've started using for homework:

  • Ask what the AI actually does — first-pass feedback, or a finished product?
  • Ask who stays accountable for the final judgement — a person, or the algorithm?
  • Ask whether it's built to scale responsibly, or whether it's optimised to be used as much as possible.

For more on how UNESCO's global guidance translates into practical questions for choosing tools, see our vetting checklist or the complete Digital Learning Week 2026 FAQ.

FAQ

What is Redação Paraná?

Redação Paraná is an AI writing-feedback programme run by the State Secretariat of Education of Paraná, a Brazilian state government, deployed across its public-school system. "Redação" is Portuguese for essay or composition. It shared UNESCO's 2026 ICT in Education Prize.

Does AI grade the students' essays instead of a teacher?

UNESCO's announcement doesn't specify the exact grading workflow, so we can't say for certain. But programmes of this kind, run by an accountable public education authority, typically use AI to give a fast first pass of feedback while a human teacher remains responsible for the final assessment — not as a full replacement for teacher marking.

Why does a state-run programme matter more than a private writing app?

Because it's accountable to a public education system rather than to subscription revenue, and it operates at genuine classroom scale rather than as an optional extra — both of which are useful signals when judging whether an AI tool is built to teach well, not just to be used a lot.

Duke Harewood built aitutors.me for his own KS3 daughter — a tutor built around the same principle Redação Paraná appears to embody: AI sharpens the work, a real accountable process decides what it means. Published 11 September 2026.