Issue 9
Florida attorney general seeks emergency court order to bar minors from ChatGPT and restrict OpenAI
Florida's attorney general is asking a court to bar minors from ChatGPT and to restrict OpenAI and its chief executive, Sam Altman, while the child-harm lawsuit proceeds. The filing, submitted on Monday, seeks an injunction covering six categories of conduct and, according to Reuters, would also affect the development of new models.
Who is affected and how
- student
- If the court grants the request, minors could lose access to ChatGPT and would need to find another tool for homework.
- parent
- Parents may want to check which of their children use ChatGPT and what to expect if the service introduces age verification.
- education policymaker
- The case could set a precedent for court-ordered limits on minors' access to generative AI, so its outcome is worth following.
- edtech company
- Products for minors that are built on ChatGPT should factor in access and compliance risk.
Harvard experiment: AI physics tutor taught more than in-class active learning
In a randomized Harvard study, students learning physics with a custom AI tutor reportedly learned more than twice as much as those in classroom instruction, in 49 minutes. Several international and…
university studentA well-designed, subject-specific AI tutor can measurably improve learning efficiency, compared with a general-purpose chatbot.teacherThe experiment compared the AI tutor with in-class active learning, so teachers may want to look at where it could supplement practice in their own subject.education policymakerThe result supports backing pilots of subject-specific AI tutors, though a single experiment cannot be generalized.edtech companyThe result argues for purpose-built tutor bots tied to course material over general-purpose chatbots.AI companion · globalMore children turn to AI for companionship, and research suggests it may leave them lonelier
A North Carolina university study found that up to a fifth of US middle schoolers use an AI chatbot as an emotional companion. Vox reports this makes children lonelier, while South Africa's News24…
parentAsk your child whether they use a chatbot as a conversation partner and what they ask of it.teacherAlongside school-based psychological warning signs, watch for students choosing a chatbot over social relationships.school leaderSchool welfare and psychology protocols should be updated to cover the issue of AI companions.studentAn AI companion is no substitute for real relationships, and the articles say it may increase loneliness.survey · United StatesReports link falling test scores to Google's school AI push and teens' AI use
Reports say 70% of teenagers use AI for schoolwork while test scores decline, and that even students consider Google's push of AI into schools to have gone too far. In a survey described in Hungarian…
studentThe articles say dependence slows thinking, so it is worth solving the harder part of an assignment yourself.teacherDesign assignments so that AI does not take over the thinking.parentAsk your children what schoolwork they use chatbots for.education policymakerCausation is not proven, but given the pattern it is worth measuring the relationship between AI use and performance.regulation · European UnionEU Kids Act would target addictive apps to protect children
The European Commission has proposed a single EU set of rules on how children may use social media and video platforms. The proposal targets addictive apps.
parentThe EU rules could bring limits on children's use of social media and video platforms.education policymakerThe EU framework may also affect national child-protection and education rules.edtech companyPlatforms that reach children need to prepare for the EU requirements.Hungary
More key stories of the week
Dartmouth authorizes Pangram AI detector as doubts over false positives persist
Dartmouth's dean of faculty has authorized professors to use the AI detector Pangram to review student work. Meanwhile, Stanford research is circulating again, finding that the detectors it tested…
university studentAn AI suspicion can be wrong, so keep your drafts and traces of your writing process.teacherA detector flag is not proof, especially for non-native writers; back it up with other evidence.school leaderRules on detector use need an appeals and evidence procedure.edtech companyTransparent reporting of false-positive rates can be a competitive advantage.survey · United StatesReport: AI suspicions are eroding trust between students and teachers
A new report examines how AI is affecting trust between students and teachers. Another article says suspicion over AI use is also undermining student-teacher relationships.
teacherClarify AI rules before assessments, because mere suspicion can damage the relationship.studentAgree on AI use with the teacher in advance, so that your work does not come under suspicion.school leaderConsistent, well-communicated AI rules can reduce trust conflicts in schools.tool · globalGoogle's Gemini study tools extend to younger students and Workspace for Education
Google has opened Gemini's study tools to younger students, and the Gemini student hub has become available to Workspace for Education users. One analysis says that from 10 August the Gemini tab in…
school leaderDecide whether to keep, restrict or replace Gemini, as it may be present in Classroom by default.teacherCheck what students can access in Classroom and adjust assignments accordingly.parentAI may already be available in your children's school accounts; ask the school.edtech companyGoogle's built-in tools narrow the market for standalone AI learning tools.survey · United StatesUS schools experiment with AI with little evidence or policy to guide them
Several US reports say schools' AI experiments are running ahead of research evidence and regulation. A national district-level report also found gaps in AI guidance, practice and support for…
school leaderLaunch AI adoption with measurable goals and evaluation, because little external evidence is available.teacherTeachers need training and rules, because in many places experiments run without institutional guidance.education policymakerWith research evidence and regulation lacking, tracking pilots is important.regulation · United StatesPennsylvania House passes bill requiring parental consent for children's use of AI companions
The Pennsylvania House voted 133–70 to pass a bill that would require parental consent for minors to use AI companions. Thirty Republicans also voted for the bill; the House is separately considering…
parentIf the bill becomes law, parental consent will be needed for a child to use an AI companion, so it is worth reviewing the apps your child uses now.edtech companyChatbots accessible to minors will need a parental-consent process, though critics complain the measure is invasive.education policymakerThe bipartisan vote suggests that regulating AI companions is a politically viable path at state level.researchAustralian research: AI now performs much better in law school exams
Researchers tested AI on real Australian law exams three years ago, when it performed poorly. According to an article in The Conversation, they have now re-examined the question and write about how…
teacherTry exam tasks with AI and redesign assessment where necessary.education policymakerThe value of traditional take-home or open-book exam formats needs to be reconsidered.university studentExams may change, as AI performance is getting closer to human level.regulation · United StatesBipartisan US bill would protect children from human-like chatbots
Rep. Miller-Meeks has introduced bipartisan legislation to protect children from human-like chatbots. The available excerpt gives no further detail on the text.
parentThe bill promises federal-level protection against human-like chatbots, so it is worth following its text.edtech companyProducts for children that communicate in a human-like way may face federal restrictions.education policymakerA federal framework may emerge alongside state rules on children's chatbot use.toolAnthropic brings Claude lesson planning to 68,000 teachers in Ghana
Anthropic is rolling out a Claude-based lesson-planning tool to 68,000 teachers in Ghana. ETIH EdTech News reported the announcement.
teacherAn AI assistant available to Ghanaian teachers could reduce the time they spend on lesson planning.education policymakerThe programme is one example of large-scale AI support for teachers.edtech companyMajor AI companies are targeting teachers directly through government programmes.child safety · United StatesDeepfakes at schools overwhelmingly target girls and women, Cybernews reports
According to Cybernews, the victims of school deepfakes are overwhelmingly girls and women. An Oregon school district is separately dealing with AI-generated social media posts targeting students and…
school leaderSchools need a set procedure for deepfake incidents, from reporting to support.teacherTeachers need to recognise the signs and know whom an affected student can turn to.studentA student who creates or shares a deepfake can be held responsible, and the target can ask the school for help.parentTalk with your children about how to react to fake images made of them.regulation · United StatesNew York City and Montclair schools restrict AI and technology use
According to the articles, edtech purchases in New York City are on pause as schools wrestle with AI, and the rules of the AI ban are being summarised for parents too. In Montclair, New Jersey…
school leaderCompare the patterns of purchase pauses and usage restrictions with your own rules.teacherCheck which AI tools may be used in class.parentAsk for the school's AI rules and align home use with them.edtech companyPaused purchases may slow sales in large US city districts.regulation · United StatesCalifornia enacts child-safety requirements for companion chatbots
California has enacted a law setting child-safety requirements for companion chatbots. According to a Morgan Lewis analysis, from 1 July 2027 providers that allow children to use their services must…
edtech companyPrepare for the child-safety measures by 1 July 2027 if your product offers companion-chatbot features to minors.parentThe law sets a mandatory level of protection for companion chatbots used by children, which providers can be held to.education policymakerCalifornia's rules could serve as a model for regulating companion chatbots in schools.survey · United StatesInstructure's first higher-education EdTech Top 40 puts dedicated AI tools outside the top 100
Instructure has launched its first EdTech Top 40 list for higher education. It found that dedicated AI tools rank outside the top 100.
university studentAmong the tools students actually use, dedicated AI tools are not the most popular.education policymakerProcurement should be based on actual usage data.edtech companyDedicated AI products still have to establish a presence in higher education.