Issue 6

Sep 7–13, 2026 · 42 stories · 9 of them from Hungary

Viewpoint

OECD: students who use AI chatbots for schoolwork score lower in PISA

According to the OECD's PISA 2025 report, students who did not use chatbots for schoolwork scored higher, while moderate users scored similarly to non-users (Fortune). Reading and maths results are declining worldwide; the founder of PISA says AI could "hollow out" young people's skills.

Sep 8, 2026 · 35 sources

Who is affected and how

student: Results differ between moderate, deliberate use and use that replaces thinking.
teacher: Tasks should be designed so that students do not use AI to replace their own thinking.
school leader: School leaders can cite the OECD data when drafting institutional AI policies.
parent: At home, parents should have children treat chatbots as a supplement to learning, not a tool for getting answers.
education policymaker: The PISA data are a main source of arguments for AI regulation and curriculum direction.
edtech company: Products must show that they support learning rather than replace it.

regulation · United States

New York City bans student use of generative AI through eighth grade for at least a year

From the start of the school year, New York City's public schools (about 800,000 students) are banning students' use of generative AI from kindergarten through eighth grade, for at least one year…

studentStudents in NYC public schools may not use AI tools for schoolwork through eighth grade.teacherTeachers must plan lessons and check compliance with the ban under the new rules, without detailed practical guidance.school leaderSchool leaders must work out local enforcement and assessment at the start of the school year.parentParents can expect schools to ask them to keep AI and screen-time limits at home too, and can take part in the review running until April 2027.education policymakerPolicymakers can treat the ban as a model or a cautionary example; there is no data yet on its impact.edtech companyEd-tech companies cannot sell generative AI aimed at students in the NYC K–8 market for at least a year.regulation · United States

Newsom signs strictest US child-safety chatbot and social media laws

California Governor Gavin Newsom on Thursday signed child-safety laws covering social media and AI chatbots, including a rule backed by Sam Altman and "Adam's Law". Pinterest welcomed the draft.

studentCalifornian teens will be covered by stricter chatbot and social media protections.parentNew legal protection, and the ability to hold platforms accountable for a child's chatbot use.education policymakerA model regulation that other states and the EU may watch.edtech companyCompliance of chatbot features aimed at young people needs to be reviewed.school leaderThe rules of the platforms students use are changing; this is worth addressing in digital safety education.school practice · United States

More US school districts restrict or ban student use of AI

The largest US districts, including New York and Los Angeles, are restricting students' use of generative AI as the school year begins, and Madison has blocked Gemini on student computers. In…

studentStudents in other districts can also expect blocked tools and stricter rules.teacherTeachers need to clarify in advance which AI tools are allowed in class and for homework.school leaderSchool leaders are responsible for the technical enforcement of bans (blocking) and for communicating the rules.parentParents should ask the school what AI rules apply to their child.education policymakerIn the debate over bans versus responsible use, district decisions set precedents.edtech companySales risk: districts may blacklist or exclude products.AI companion · United States

Teens increasingly turn to AI chatbots for mental health support, and schools are adopting them

According to EdSource, teens increasingly turn to chatbots for emotional support, while schools introduce chatbots amid a counselor shortage; the Kiwi chatbot helps detect suicide plans, but safety…

studentStudents often turn to AI with emotional difficulties, but it is no substitute for a professional.teacherTeachers need to recognise warning signs and know where to direct a student in distress.school leaderSchool leaders adopting a chatbot must settle data privacy, the escalation protocol and accountability.parentParents are advised to talk with their child about how they use chatbots for emotional questions.edtech companyFor mental health features, demonstrating safety and escalation mechanisms is expected.university studentAI used for emotional support can complement university counselling services, but not replace them.

Hungary

surveyPISA 2025: Hungarian results at historic low, about a third of students use AI for writingstudentAbout a third of Hungarian students are among those who use AI for written tasks.teacherThe forthcoming AI guide and strategy will set the framework for classroom use.school leaderThe results put pressure on schools to rethink institutional AI rules and skills development.parentGiven the results, parents should pay attention to what their child uses AI for in homework.education policymakerDeveloping a national AI strategy and guide is now on the agenda.edtech companyIn the Hungarian education AI market, the framework of the guide and strategy will be decisive.school practiceHungarian debate: critical thinking first, then AI in schoolsteacherIn the Hungarian debate, most teachers ask for a slower, more deliberate introduction.school leaderSchool leaders can decide in what order to introduce AI education and critical-thinking development.education policymakerPolicymakers preparing regulation must also consider teachers' concerns about the pace.parentParents should ask the school how AI use is taught.AI companionHungarian experts: teens choose AI girlfriends, and the age limit for chatbot use is debatedparentExperts recommend limiting a child's solo chatbot use and talking about it together.studentFor sharing confidential problems, a chatbot is no substitute for a person.teacherTeachers should watch for withdrawn students and for signs of attachment to AI companions.school leaderThe topic can be included in school psychology and digital education programmes.researchELTE study identifies what determines whether students keep using ChatGPTuniversity studentThe study shows which factors students weigh when deciding whether to use it.teacherThe results can inform instructors' decisions on AI integration.school leaderInstitutional AI strategy should be aligned with the findings.edtech companyFactors in student acceptance are useful for product development.school practiceÓbuda University starts the year with an AI-native strategy and plans an AI bachelor's programmeuniversity studentA new AI programme and an AI-native learning environment are expected.teacherInstructors will have to adapt to AI-integrated course material.education policymakerA Hungarian example of accreditation and programme-launch decisions.school leaderProvides a strategic model for universities.school practiceSlovakia launches school AI programme; Magyar Iskola issues guide for Hungarian-language teachersteacherPractical considerations have been compiled for Hungarian-language teachers in Slovakia.education policymakerSlovakia's school AI programme is a regional benchmark.regulationWould Hungary need a New York-style AI ban? Domestic debate beginseducation policymakerOn the question of a ban or regulation in Hungary, policymakers must take a position based on foreign examples.school leaderSchool leaders must draw up institutional rules until there is a central one.parentThe outcome of the Hungarian ban debate will determine what parents can expect from schools.surveyNearly half of Polish teachers use AI; Copernicus Science Centre launches free courseteacherA free course is available on using AI in teaching.education policymakerThe share offers a benchmark for teachers' AI use in the region.regulationUkraine prepares AI law that also covers educationeducation policymakerThe framework regulation, which also covers education, is a regional example.school leaderInstitutional compliance can be expected once the law is adopted.

More key stories of the week

regulation · United States

Microsoft and teachers unions announce AI privacy standard for schools

Microsoft, together with the AFT and the UFT, announced the "National AI Safety & Privacy Standard", a ten-point framework that can be written into contracts, covering the privacy and safety of AI…

teacherContractual protections and transparent data handling are expected for school AI tools.school leaderThe standard can be built into procurement contracts as a benchmark for AI tool privacy.parentParents can ask the school specific questions about how their child's data is handled.education policymakerIt offers a practical template for privacy requirements in AI procurement.edtech companyOther vendors may be pushed to offer similar guarantees.AI companion · global

Global study: a third of young people have an AI companion; minors are a large share of character chat app users

A global survey finds a third of young people have an AI companion while close friendships narrow. Other reports say minors make up a large share of AI character chat app users, and concern over…

studentAn AI companion can create emotional dependence that may crowd out real relationships.parentParents should keep an eye on their child's use of character chat apps and the time spent on them.teacherWeakening ties between students may show up in the classroom community.education policymakerAge limits and regulation for AI companions are now coming onto the agenda.edtech companyFor products reaching young users, engagement patterns and age verification may come under scrutiny.tool · global

OpenAI launches ChatGPT for Teens with parental controls and study mode

OpenAI has launched a ChatGPT for Teens mode with stronger safeguards and parental controls, promoted mainly as an educational tool. Hungarian-language guides on setting up parental controls have…

parentParents can set up supervision and content filtering on a teen's account.studentIn teen mode, students get filtered content and learning features.teacherTeachers should know the safeguards and learning features of a tool their students use.edtech companyOpenAI's entry into the education segment creates competitive pressure.tool · United States

Google and OpenAI target US teachers with free AI training and tools

Google is expanding its free AI Educator Series with monthly Gemini modules for 6 million US teachers, Missouri's education agencies are launching free AI training with Google, and Northside ISD is…

teacherFree training and tools are available, but the risk of dependence on a company should be weighed.school leaderDistrict-level AI adoption requires evaluating agreements with companies.edtech companyFree offerings from large platforms put pressure on smaller players.education policymakerVendor dependence must be considered in state training programmes.tool · global

AI detectors prove unreliable after paraphrasing; watermarking raises concerns for non-native writers

Reports say AI detectors become unreliable after paraphrasing, and a study found that texts by non-native writers are wrongly flagged as AI-generated. Watermarking is also being tried, but it does…

teacherA detector result alone cannot serve as evidence of cheating.university studentStudents face false suspicion of AI use, especially when writing in a non-native language.school leaderProcedures should also record the limits of detectors.edtech companyDetection and watermarking products must manage the risk of false flags.school practice · global

Teachers and professors shift exams toward proving students are learning as AI glasses pass proctored tests

University professors and teachers are shifting the emphasis towards making cheating harder, verifying that students have learned and assessing the learning process; AI glasses can already pass…

teacherShould move to oral, process-based or in-person assessment.university studentProving their own workflow may matter more in assessment.school leaderExam-security procedures need to be adapted to AI tools.edtech companyDemand is growing for exam-security and proctoring tools.case · Asia

18 months in jail for 1,100 deepfakes; South Korean courts punish student deepfake abuse

A student received 18 months in jail for creating 1,100 deepfakes of lab colleagues (Korea Herald). South Korean courts punish students' deepfake abuse based on intent and harm.

studentCreating deepfakes can carry serious criminal consequences.teacherTeachers must recognise and report such abuse.school leaderSchool leaders need a clear procedure for handling deepfake cases.university studentDeepfakes made on campus can be punished with prison.parentParents should talk with their child about the consequences of deepfakes.regulation · global

UNESCO issues discussion paper on sustaining education as a common good in the age of AI

Ahead of Digital Learning Week 2026, UNESCO published a discussion paper titled "Sustaining education as a common good in the age of AI: The case for deliberative governance" and launched a global…

education policymakerThe paper proposes AI governance based on public consultation.school leaderThe global guidelines could serve as a starting point for institutional AI rules.teacherThe consultation may also involve teachers.tool

UNSW gives 80,000 students and staff ChatGPT Edu and adds new academic integrity measures

An agreement between OpenAI and UNSW Sydney gives 80,000 students and staff access to ChatGPT Edu. The university is introducing new administrative measures to deal with AI cheating.

university studentStudents get official AI access, but under stricter integrity rules.teacherAssessment must be designed with AI use in mind.school leaderAn example of managing university AI adoption and integrity together.edtech companyOpenAI's growing presence at universities also affects its competitors.regulation · United States

US House education committee investigates how AI affects college admissions

The House education committee is investigating how AI affects college admissions. Meanwhile, admissions offices are saying how they handle essays written with AI.

studentUsing AI in an admissions essay carries an authenticity risk.university studentThe admissions process may change as a result of AI.education policymakerThe committee's investigation could lead to regulatory steps.school leaderCounsellors need to clarify the limits of AI use with students.survey · global

Surveys: one in five child AI users report potential harm

Australia's eSafety report finds that one in five child AI users report potential harm. Research by Terre des Hommes Netherlands finds displaced Ukrainian children use AI every day but overestimate…

parentRegular conversation about a child's AI use matters more than relying on platform safeguards.teacherAI safety is worth teaching as part of digital education.education policymakerThe surveys can support the case for AI regulation and safety requirements.child safety · United Kingdom

Deepfakes sexualising students spread through British schools

The Independent reports on an epidemic of sexualised deepfakes of students spreading through British schools.

studentA student can be both a victim and a perpetrator of deepfakes; a victim needs to ask for help.teacherTeachers need to be prepared to handle and report such cases.school leaderSchool leaders need to develop procedures and a victim-protection protocol.parentParents need to talk with their child about the dangers of image manipulation.

In brief

← Issue 5 · Issue 7 →