The International Baccalaureate's Global Academic-Integrity Guidance on AI Use
Switzerland
In February 2026 the IB updated its global Academic Integrity Policy with a new AI appendix, distinguishing acceptable AI-assisted learning …
Brazil · Curitiba · See the Brazil profile · See the Curitiba profile
Top 25% 60/100 · Ask Evidence Copilot about this practice
Paraná's state education department runs Redação Paraná, a writing platform with 833,000+ active users; AI correction (Google technology) gives instant feedback, with teachers handling subjective aspects. Reported scale is large; learning-impact evidence is still descriptive.
Redação Paraná is a writing platform launched in 2021 by the Paraná State Education Department (Seed-PR) for students in the state school network. According to CONSED, it has more than 833,000 active users, and in 2025 more than 4.6 million essays were completed on it.
AI correction, developed with Google, was first applied to argumentative essays (dissertativo-argumentativo) in 2025 and has been extended to short story, chronicle and narrative genres. Reported criteria include fit to genre, organisation and progression of ideas, theme development, and language use. Approximately 153,000 texts were AI-corrected in 2025. Teachers remain responsible for discursive and subjective aspects; an academic article in Texto Livre analyses this division of correction between AI and teacher.
Evidence caveat: published figures are usage counts and an anecdotal teacher time saving (about five minutes per essay versus seconds). No controlled evaluation of the Paraná platform was found. A randomised trial of a comparable AI essay-feedback system in Espírito Santo (J-PAL) found +0.09 SD in essay scores, which suggests but does not demonstrate similar effects here.
Read the full analysis: https://www.consed.org.br/noticia/redacao-parana-correcao-por-inteligencia-artificial-e-ampliada-para-novos-generos-textuais
Implementation detail (cost, timeline, staffing, conditions for success) is not yet available for this practice.
Do you run this practice? Claim it — verified implementers get a public contact pathway and can propose corrections.
Where this practice's information was retrieved from, and when.
Switzerland
In February 2026 the IB updated its global Academic Integrity Policy with a new AI appendix, distinguishing acceptable AI-assisted learning …
Dominican Republic
SARA pairs a validated Spanish reading battery with AI-automated scoring on a tablet. Validated on 1,860 Santo Domingo pupils, its …
United States of America
Vanderbilt disabled Turnitin's AI-writing detector campus-wide in Aug 2023, calculating its claimed 1% false-positive rate could wrongly flag ~750 of …
Germany
Mühlhoff and Henningsen (Osnabrück) tested Fobizz's AI grading tool in two test series: scores and feedback looked arbitrary, one text …
Open full copilot Grounded in cited practices — always check the sources.