Ethical AI in education, built by and teachers
Every partner NILO backs sits where our two areas of work meet: the inner capacities that let people and communities adapt, and the wider transition to an economy that protects people and planet. Few partners hold the first of those as directly as EcolHuma.
For more than a decade, the French charity has worked from one stubborn conviction: you cannot improve a school system without first supporting the humans who run it every day. This year, that conviction has taken EcolHuma somewhere new — into one of the most consequential and least regulated technology shifts hitting education anywhere in the world.
AI is reaching classrooms faster than any shared understanding of how it should be used. For teachers, the shift is arriving largely without prior notification to them. Tools designed by private tech companies, trained outside real classrooms, and handed down with little say in how they work.
From a peer platform to AI in education
EcolHuma was founded in 2012 by Florence Rizzo. Her own path — a childhood in rural eastern France, a family that put education front and center, and a career across Sciences Po, Ashoka and international policy — shaped an organisation with a plain premise: teachers are not the obstacle to a better school system. They are its precondition. Its flagship platform, ÊtrePROF, launched in 2017 and now reaches close to 200,000 teachers — roughly 1 in 4 in France — with peer support, practical training and a growing bank of pedagogical resources.
In 2025, EcolHuma made a deliberate turn: to build AI into ÊtrePROF rather than watch the market build it around teachers instead. A beta launched in 2025 with the platform’s first AI tools, including a lesson-planning generator that draws on generative AI and a pedagogical coach. Both were tested from the outset with more than 500 teacher “builders”, who shaped how the tools worked before any wider rollout. The assistant now walks a teacher through their profile, a specific class’s needs — learning differences, group dynamics and classroom atmosphere — and a full lesson sequence grounded in the official curriculum and in everyday educational practice.
Why the ethical use of AI matters as much as the technology
EcolHuma’s own research explains why this was necessary rather than optional. Its 2025 barometer, conducted with 1,962 French teachers, found a workforce that is neither uncritically enthusiastic nor reflexively hostile. 41% see AI as a genuine efficiency tool. 38% see it as a powerful pedagogical lever that has to be handled with real care. 21% see it mainly as a worry.
What stands out is the scale of unmet need behind those numbers: 78% wanted proper training on AI, and 59% wanted clearer rules of use, well ahead of any settled view on what “good use” even looks like.
That gap is where the commercial EdTech sector is moving fastest, and not always towards teachers’ interests. Much of the AI now reaching classrooms arrives opaque, commercially driven and untested in real settings. The risks of AI in that form are not abstract: bias baked into tools no one can inspect, AI systems optimised for engagement rather than learning, and a widening gap between classrooms that can navigate the technology and those left to absorb whatever they are given.
Handled well, the same tools could be more equitable, not less. The difference lies in the ethical considerations built in from the start, and in who gets to weigh them.
Ethical AI, built with teachers rather than sold to them
Florence Rizzo has been direct about what this means. Teachers work in professions of connection, care and craft that AI cannot replace. Treating education as a cost rather than a shared investment is precisely the mistake a purely commercial rollout risks making. EcolHuma’s answer is not to resist AI, but to decide how to integrate AI responsibly and on teachers’ terms.
The assistant is shaped by practising teachers and tested in real classrooms, from nursery through to the final year of secondary. It answers to a scientific committee that includes a cognitive psychologist and a UNESCO chair holder on AI in education, rather than optimised primarily for engagement metrics or subscription revenue. The ethical principles are not a statement bolted on afterwards; they are the framework the tool is built around.
What NILO sees in this use of AI in education
This is the kind of work NILO’s education and social-emotional learning theme exists to fund. Not because teacher wellbeing and AI ethics sound urgent in the abstract, but because the people doing the daily work of social-emotional development — building the empathy, critical thinking and human connection the next generation will need — are themselves under-resourced, under-trained, and now facing technology decisions made largely without them.
EcolHuma is making sure teachers, not the private sector, define what a good assistant for teaching and learning actually looks like, and meeting the ethical challenges of that head-on rather than leaving them to the market. Philanthropy has its own version of the same temptation: to fund the visible tool and forget the humans using it. The harder, slower task is ensuring ethical use is decided by the people closest to the classroom.