Issue 7
Leaked EU Kids Act plan would restrict social media, games and AI chatbots for under-15s
According to leaked documents, the European Commission would restrict access to social media, online games and AI chatbots for children under 15, and it formally presented the draft EU Kids Act in the middle of the month. The text mentions a ban on social media without parental consent and "safe by design" rules, with fines that could reach 6% of global revenue.
Who is affected and how
- parent
- Parents should expect age verification and parental consent to be required for a child's chatbot and social media accounts.
- student
- Under the draft, those under 15 could lose free access to chatbots and online games.
- education policymaker
- Member states' school AI rules will need to be aligned with the forthcoming EU framework, and its transatlantic effects are already being debated.
- edtech company
- AI products that reach children must be designed with age verification and safe defaults, given the significant risk of fines.
OECD: using AI for schoolwork leaves students 1.5 years behind, PISA 2025 finds
According to Bloomberg, the OECD found that students who use AI for schoolwork fall 1.5 years behind. PISA 2025 data show, for example, that 41% of students in Thailand used AI for preliminary…
education policymakerBased on the PISA data, policymakers need to regulate the difference between AI that supports learning and AI that replaces the task.teacherTeachers should design homework and assignments so that AI does not take over the thinking.studentUsing AI as a tutor versus as a shortcut can lead to different learning outcomes.parentParents should ask whether their child uses the chatbot to get explanations or to have the task solved.school practice · United StatesNew York City and Los Angeles schools suspend student-facing AI use
New York City's public schools are temporarily pausing AI use for students from pre-K to grade 8 for 2026–27, and Los Angeles has likewise introduced a ban on student-facing AI tools. According to…
teacherWith student-facing AI tools banned, rules for AI use in teachers' planning and administration need to be clarified separately.school leaderThe school policy and parent information should state clearly which tools are allowed for homework.parentAlongside a school ban, AI use at home needs a separate decision; parent groups say a ban alone is not enough.edtech companyLarge districts' bans directly shrink the market for student-facing products; industry arguments stress the hidden cost of a ban.education policymakerThe New York and Los Angeles examples give other districts a model for their decisions.survey · United StatesNearly one in five young people turn to AI for mental health advice, reports say
A report from Utah says nearly one in five young people turn to AI for mental health advice, while medical trade publications write about what doctors should know. A systematic review examined the…
parentIt is worth asking your child whether they use AI for emotional support, and where the human channels for seeking help are.teacherSchool mental health work has to take into account that students turn to chatbots.school leaderInformation materials on student mental health care should also flag the risks of AI chatbots.regulation · United StatesCalifornia signs child-safety laws restricting AI chatbots and social media for teens
California's governor has signed more than a dozen laws regulating teenagers' use of social media and chatbots, including Adam's Law, named after a teenager who died by suicide after conversations…
parentThe California model points toward parental notification and age verification, so it is worth learning the platforms' new settings.education policymakerThe package could serve as a template for other regulators on chatbot obligations towards minors.edtech companyChatbots aimed at minors must provide age verification and protective features.studentTeenagers' chatbot and social media use across California will face stricter limits.Hungary
More key stories of the week
MIT report warns AI causes 'cognitive surrender' among US college students
An MIT panel report says AI use weakens students' mental effort, with the authors describing it as "cognitive surrender". The topic appeared in several media outlets in mid-September, alongside…
university studentThe report highlights the risk of students becoming dependent on AI, so they should start an assignment on their own before turning to it.teacherTeachers should shape assessment so that it measures actual knowledge.parentA child's AI use for homework may replace thinking, so parents should watch how it is used.AI companion · European UnionEU Kids Act draft would strip companion chatbots of emotion and memory for under-18s
Under the draft, companion chatbots could not show emotions or remember earlier conversations with users under 18. Analyses say this would remove the very feature that makes Character.AI appealing to…
parentThe teen version of AI companion apps may be noticeably more limited, so it is worth talking to the child about this in advance.studentPopular companion apps may lose their personal feel in teen mode.edtech companyCompanion-style AI products for under-18s will need their emotion and memory features redesigned.tool · globalUniversities roll out ChatGPT Edu: Sydney, Maine, Utah and SUTD
The University of Sydney is introducing ChatGPT Edu for students, while in Maine the rollout comes with open questions over funding. The University of Utah clarified how it communicated a cost change…
university studentAn institutional licence can give free access, but the funding may change.education policymakerSustainable funding for a licence needs to be planned in advance.edtech companyInstitutional licences are shaping the competitive landscape for AI tools aimed at students.case · AsiaIIT Bombay student caught using ChatGPT in an exam dies by suicide
A student at IIT Bombay was caught using ChatGPT during an exam and later died by suicide. The institute says no disciplinary action was taken in the case.
university studentPressure over AI cheating allegations can cause mental strain, and support services may be needed.teacherWhen handling suspected AI use, the protection of the student concerned is also a consideration.education policymakerIntegrity procedures and student support have to be planned together.tool · United StatesMicrosoft makes AI privacy and safeguard commitments for schools amid growing criticism
Microsoft has committed to sweeping AI privacy rules for schools and students amid growing criticism. The open question is whether other big tech companies will follow.
school leaderIn light of Microsoft's commitments, schools should compare their data-processing contracts with those of other vendors.edtech companyCompetitors will need to show a similar level of student data protection commitments in procurement.parentThe commitments give parents a reference point for questions to ask the school about how student data is handled.research · United StatesResearch on AI tutors: targeted instruction is promising, but chatbots can reduce student effort
A new analysis finds that AI learning tools are promising for targeted instruction, but chatbots can reduce student effort. In a study of 6,000 students, the AI tutor helped when it slowed students…
teacherAn AI tutor is effective when it forces the student to think through their mistake.edtech companyDesign that slows learning and has students work through their errors is the more effective direction.education policymakerEvidence on the educational benefits of AI is disputed, so it is worth asking for independent data before procurement.regulation · United StatesFlorida adopts statewide AI rules for schools from pre-K through college
Florida's Board of Education on September 16 approved new rules governing the use of AI and education technology in schools, from voluntary pre-kindergarten (VPK) through higher education. The rules…
teacherTeachers need to know what they may use AI for in class and what they can allow students to do.studentAI use in schools and colleges will follow the new rules.school leaderSchool policies will have to be aligned with the state rule.education policymakerFlorida is positioning its regulation as a national model.tool · globalGoogle opens Gemini to all students as schools still catch up
Google's Gemini is now available to all students, while schools are still preparing for the new situation. In Ukraine, students get Gemini free for a year, and parents are being given explanations of…
teacherTeachers need to clarify which Gemini features are permitted in the classroom and how they appear in assignments.school leaderSchool leaders should decide student access settings together with the school's Workspace administrator.parentGemini behind a school account works differently from one on a personal account, so parents should ask about it.studentThe tool is free to use, but it must be used in line with school rules.regulation · United StatesUS states pass chatbot safety laws one after another, with Google helping to draft the text
US states have numerous laws and bills aimed at the risks chatbots pose to children and teenagers. According to NPR, Google is involved in drafting chatbot laws across the country, and in Oklahoma…
education policymakerRegulatory templates and industry lobbying are both present, so sources are worth checking when drafting state legislation.edtech companyThe patchwork of state rules means differing compliance obligations in the US market.parentChild protection currently rests on legal frameworks that differ from state to state.school practice · globalUniversities in South Korea, Dublin and Norway act against AI cheating
South Korean universities are tightening action against AI cheating in exams, and University College Dublin (UCD) recorded almost 50 cases of unauthorised AI use last year. Uncontrolled AI use is…
university studentAI use in exams and assignments must be disclosed as the rules require, and detectors can make errors.teacherSuspicion should not rest solely on detectors, given their false positives.education policymakerIntegrity rules should be designed with detector errors in mind.tool · globalGates Foundation commits $1 billion to AI, with $400 million earmarked for schools
The Gates Foundation has committed at least $1 billion to AI in health, agriculture and education. Teachers are warning about the $400 million earmarked for AI in schools.
teacherTeachers' warnings suggest that teacher perspectives should be taken into account in school AI funding.education policymakerLarge foundation funds are shaping school AI priorities.edtech companyA significant funding source is opening up for school AI solutions.child safety · United StatesCommon Sense Media: Perplexity AI search poses unacceptable risk to kids
A Common Sense Media assessment finds that the Perplexity AI search engine poses an unacceptable risk to children. The assessment was published on 15 September.
parentParents should reconsider whether their child uses Perplexity.teacherWhen recommending search tools to students, teachers should take Common Sense's rating into account.edtech companySearch tools that reach children need to pass a safety assessment.