Fifty-Two Percent of Portugal's Children Ran Into Online Hate Speech Last Year, and 85 Percent Have Already Used Generative AI

A new national report on how Portuguese children use the internet landed on Wednesday, and it contains two numbers that will be quoted separately but belong together. Just over half of the children and young people surveyed, 52 percent, say they saw hate messages or comments online at least once in the past year. And 85 percent of the same group, aged nine to seventeen, have already used generative artificial intelligence.

The study, titled "Novos usos, novos riscos? Direitos digitais de crianças e jovens (9-17 anos)" ("New uses, new risks? Digital rights of children and young people, 9 to 17"), forms part of the EU Kids Online 2025 round and surveyed 2,111 young internet users across internet use, digital skills and literacy, digital rights, risks, and the way families and schools mediate all of it.

The exposure numbers

• 52 percent saw hate messages or comments online at least once in the past year. Exposure climbs sharply with age: roughly a quarter of nine to eleven year olds, and 72 percent of sixteen and seventeen year olds.
• The content in question targets people or groups on grounds including gender, ethnic origin, religion, nationality, sexual orientation and disability.
• Boys and girls differ little on overall exposure. Socioeconomic status splits it more: 45 percent among young people of low socioeconomic status, 56 percent at moderate level and 52 percent at high level.
• 37 percent encountered conspiracy theories in the past year, rising to 47 percent among sixteen and seventeen year olds.
• The researchers also logged contact with violent or disgusting imagery, adult pornography, challenges capable of causing serious injury, drug-related material and messages about self-harm.

The finding the authors press hardest is that a substantial share of this exposure is not sought. Children are not going looking for it; it arrives. That shifts the weight of responsibility from the child to the platform, and specifically to recommender systems. Cristina Ponte, a researcher at the Instituto de Comunicação da NOVA (ICNOVA) and one of the report's authors, tied the rise in hostile content to the economics of engagement, telling Lusa that the growing dramatisation of emotions "feeds the algorithms" and that all of it is now bound up with artificial intelligence, something that has intensified sharply in recent times.

Artificial intelligence arrived without the adults

The AI figures are the most striking part of the report because of how fast they moved. Eight in ten of these children, 85 percent precisely, have used generative AI tools. Some 47 percent have used them for schoolwork, and that use rises steadily with age:

• 28 percent of nine to eleven year olds use AI tools for school tasks;
• 48 percent of twelve to fourteen year olds;
• 63 percent of fifteen to seventeen year olds.

Socioeconomic status splits this too. About half of young people in high and moderate socioeconomic households use AI for schoolwork, against 34 percent of those in the lowest bracket. That is a competence gap forming in real time, and it runs the opposite way to the hate-speech exposure gap.

Ponte's framing is that the technology entered Portuguese childhood "à margem dos professores, à margem dos pais" (outside teachers, outside parents). She does not treat school use of AI as inherently a problem: it can help a student revise, summarise or have a text explained, provided there is pedagogical guidance. The risk she identifies is narrower and more precise, which is when the tool stops supporting the learning process and replaces it, with students delegating the work itself.

Two further numbers deserve attention because they are not about homework at all. Around 23 percent of these young people have used AI tools to talk about or seek advice on physical health, and the same share has taken personal worries to them. Whatever one thinks about a chatbot as a study aid, close to a quarter of Portuguese nine to seventeen year olds are now using one as a confidant.

The researchers also flag a gap between how confident children say they are about their digital skills and what they actually understand about algorithms, the business models behind social networks, and the effects of the technologies they use. About half say they would like to learn more about fake news and personal data protection, which is a fairly direct request that somebody teach them.

The phone ban nobody asked for

Portugal banned internet-capable mobile devices in the first and second cycles of basic education from September 2025, and the report captures how that has settled. Among nine to eleven year olds, 76 percent report that the rule exists at their school, which lines up with the law.

Further up the age range, the picture inverts. Two in three children and young people either disagree with a ban on bringing phones to school or hope such a rule is never applied to them. Forty percent say no such rule exists at their school and hope none arrives. Among fifteen to seventeen year olds that figure reaches 64 percent, and eight in ten of that group disagree with a ban outright.

Offered a middle position, where the phone may be brought to school but used only in an emergency or for health reasons, opinion divides almost exactly: 49 percent agree, 51 percent disagree.

Schools are doing more than they were

The one clearly positive trend in the report is the mediation data, comparing 2018 with 2025:

• Children reporting teachers who suggest how to use the internet safely rose from 33 percent to 44 percent.
• Teachers who set rules about internet use rose from 34 percent to 43 percent.
• Teachers who support students when something online upsets them rose from 16 percent to 37 percent, more than doubling.

That last line is the most useful number in the study. It is the only one that measures whether a child has somewhere to go, and it has moved further than any of the others. It is also still only 37 percent.

What this means for families in Portugal

• The exposure is not a discipline problem. If a majority of teenagers see hate speech without looking for it, the parental conversation that works is not "what were you searching for" but "what turned up, and what did you do about it". The report's own emphasis on recommender systems supports that.
• Ask about AI before the school does. With 63 percent of fifteen to seventeen year olds using generative AI for schoolwork and adults largely out of the loop, the practical question for a family is whether the tool is being used to understand the work or to avoid it. That distinction is invisible in a finished assignment.
• Treat the health and personal-advice use seriously. Roughly one in four is asking an AI system about physical health or personal worries. That is a category of use worth knowing about in a household, particularly for newly arrived families whose children may be navigating a new school system in a second language.
• Portugal's phone rule only covers the younger cycles. If your child is in secondary education, there may well be no rule at their school, and the survey suggests they would prefer it stayed that way.
• Data literacy is the gap the children themselves name. Half want to learn more about fake news and personal data protection. Our guide to your data rights in Portugal covers the one-month reply deadline and the three complaint forms the CNPD runs, which is the machinery a parent would actually use.

The wider picture

None of this arrives in isolation. Portugal's national cybersecurity centre logged 3,864 incidents in 2025, up 40 percent year on year, with internet safety line calls climbing 39 percent and 13 percent of victims children or young people. The health side of childhood is under similar scrutiny, with almost 40 percent of Portuguese children aged five to fourteen carrying excess weight, and the school system is being asked to absorb more of this work through the new national school health programme. All of it lands on a teaching workforce that is short of 2,650 timetables this month alone.

The report does not propose a fix, and it is stronger for it. What it establishes is that the two biggest changes in Portuguese childhood online, the ambient presence of hostile content and the sudden ubiquity of generative AI, both happened without the institutions around the child being consulted. The teacher-support figure doubling since 2018 suggests the schools are catching up. The 72 percent figure among sixteen and seventeen year olds suggests they have some distance left to run.