Issue 1
Mexico's UNAM orders 58,000 students to retake AI-proctored remote exam after failure
A remote exam supervised by AI at Mexico's UNAM went so badly that 58,000 students must retake the test. The case has raised questions about the reliability of AI-based proctoring and uncertainty over admissions (from August 3, 2026).
Who is affected and how
- student
- If a technical or scoring error occurs in an AI-proctored remote exam, the test must be retaken, so it is worth documenting the exam conditions.
- university student
- Applicants and exam-takers should expect false flags from AI proctoring and the need to appeal.
- school leader
- Before a high-stakes remote exam, a backup and appeals procedure needs to be in place.
- education policymaker
- For high-stakes exams, a pilot and independent review are needed before AI proctoring is introduced.
- edtech company
- The reliability and error handling of proctoring software will come under scrutiny, and questions of contractual liability arise.
Denmark requires oral defences of essays in upper-secondary schools to curb AI cheating
The Danish government is introducing immediate measures in upper-secondary schools to curb AI-written assignments: students must orally defend their essays. The government would also have several…
studentStudents must be able to defend a submitted essay orally, so they need to build knowledge behind the text, not just the text.teacherTeachers must build oral defences and more supervised, in-school written assignments into assessment.school leaderSchool leaders must reorganise assessment arrangements and timetables to fit oral defences and supervised written sessions.education policymakerThe Danish approach is a working model and an alternative to take-home essays and AI detectors; it is worth following how the rollout goes.AI companion · globalAI companions may deepen loneliness; China bans them for teenagers
A Stanford study found that AI companions may worsen loneliness among vulnerable users; another study examines the link between interaction with AI companions and well-being. One report says AI…
parentShould watch for secrecy and for the child becoming anxious when the chatbot is unavailable.studentAn AI companion is no substitute for real relationships and is especially risky for those in a vulnerable state.education policymakerThe Chinese ban and the New York law show two different regulatory models.edtech companyProducts offering AI-companion-style features face age-verification requirements and regulatory risk.survey · United StatesGrowing numbers of young people turn to AI chatbots for mental health support
A survey cited by the Harvard Gazette found that use of AI chatbots for mental health purposes among adolescents and young adults rose 60 percent in a year; another report shows the same pattern…
parentShould ask whether their child uses a chatbot for emotional problems, and should know the real channels for getting help.university studentIn a crisis, a chatbot is no substitute for a professional.school leaderSchool mental health services must take chatbot use into account.education policymakerState-level regulatory models are emerging for teenagers' use of chatbots for mental health.child safety · United KingdomRising number of UK children report seeing explicit deepfakes of themselves
A recent report says the number of UK children who have reported seeing explicit deepfakes of themselves is rising (8 August 2026).
parentParents should talk with their child about what to do if they find fake content made of them.teacherTeachers must prepare to handle deepfake cases and support students.school leaderSchool leaders need to build deepfake cases into the child protection protocol.studentA student can report a fake explicit image made of them and ask for help.Hungary
More key stories of the week
US Senate prepares package of bills on privacy, AI and children's safety
On 4 August 2026, the US Senate was set to debate a package of bills on privacy, AI and children's safety. The proposals are also criticised: the EFF writes about privacy paradoxes in the Youth AI…
parentThe planned rules may affect parental control over children's AI use and data handling.edtech companyAI rules affecting minors could bring compliance burdens, though the details are still disputed.education policymakerThe US regulatory model may affect the procurement conditions for AI tools in schools.tool · globalOpenAI adds education plugins to ChatGPT Work and Codex
On 4 August 2026, OpenAI announced new education plugins for ChatGPT Work and Codex, aimed at teachers and students. The plugins move AI from one-off answers toward multi-step workflows, while…
teacherTeachers can delegate multi-step workflows (such as planning and preparing materials), but must check the results themselves.studentManaging a process and judging AI output become more valuable skills.edtech companyMajor model developers are entering educational workflows directly, which affects product positioning.school leaderThe new features call for a review of the approved-tools list and data protection terms.research · United StatesAi2 introduces TutorMoments, a test of whether AI tutors know when to help and when to hold back
The Allen Institute for AI (Ai2) has introduced TutorMoments, which measures whether AI tutors recognise when to help and when to hold back (August 7, 2026). Coverage in Hungarian also reported on…
edtech companyA new evaluation criterion for developing tutor bots: the timing of intervention can be measured.teacherWhen choosing an AI tutor, it is worth checking when it offers help and when it lets the student think.school practice · globalAI glasses, hidden prompts, handwritten exams: universities rethink exam security
AI glasses are becoming harder to detect, universities are updating their exam rules, and handwritten exams are spreading. A Mississippi professor's hidden prompt caught 31–32 of 35 students using AI…
university studentExam rules may cover AI glasses and hidden traps, and the risk of being caught cheating is high.teacherCan choose among hidden prompts, handwritten exams and invigilation; each method has consequences.school leaderExam regulations need to name wearable devices explicitly.education policymakerEducators have called for rules on the use of AI in universities.survey · globalMore than half of children and teens use AI for personal reasons, study finds
According to a recent report, more than half of children and teenagers use artificial intelligence for personal purposes, which carries risks (August 6, 2026). The article draws attention to the…
parentParents should ask what personal purposes, not just schoolwork, their child uses AI for.teacherTeachers should expect that students use AI for more than solving assignments.studentStudents should make informed decisions about the risks of using AI for personal purposes.research · United StatesAI tutors praise instead of teaching, and students may only look like they are learning
Articles in early August say AI tutors offer too much praise instead of teaching, and students hand in work that looks solved rather than showing that they have learned. The debate is over whether AI…
teacherAssessment has to measure the process and understanding, not just the finished product.studentIf AI only affirms and does not teach, ask it for explanations and practice, not just the solution.edtech companyDemonstrable learning gains, not praise, will become the yardstick.case · United StatesFalse AI accusations: detector false positives and students' rights
A US high school student says she was falsely accused of writing a paper with AI (NBC News). Analyses say AI detectors' false positives are especially common among non-native and diligent writers…
studentKeep drafts and version history so you can prove your own work.teacherA detector flag is not evidence on its own; conversation and observation of the writing process are needed.school leaderAn appeals procedure must be provided for cases of suspected AI use.university studentA false accusation can be appealed, and documenting your workflow protects you.school practice · United StatesPalm Beach County schools move to crack down on AI-driven student bullying
The Palm Beach County school district is considering a tougher response and harsher punishments for students who use deepfakes to target others (3 August 2026).
school leaderSchool leaders must write sanctions for deepfake bullying and a reporting procedure into the code of conduct.studentManipulating other people's images with AI can lead to disciplinary consequences.parentParents should talk with their children about the consequences of deepfakes.teacherTeachers need to know where to forward a suspected case of deepfake bullying.school practice · United StatesTeacher AI training: union programme funded with $23m from big tech, plus university and school grants
A teachers' union is training teachers to use AI with $23m in funding from big tech companies (Guardian). Stanford has awarded twelve seed grants for AI in education, and Irvine schools have received…
teacherTech-funded training is becoming available; questions of independence are worth weighing.education policymakerFunding for teacher training is shifting towards tech companies, which raises questions of dependence.survey · globalNearly a third of college students use AI in daily life, but governance lags
A study finds that nearly a third of college students use AI as part of daily life. Another analysis says generative AI has become mainstream in higher education, but governance is lagging (4 August…
university studentAI is an everyday tool, and the rules on its use differ from institution to institution.school leaderThe lack of governance has to be filled with institutional policies.education policymakerAddressing the regulatory gap is urgent.tool · globalGemini in Google Classroom expands to users of all ages
According to Google Workspace Updates, Gemini in Google Classroom is expanding to users of all ages, with contextual starter prompts for students (4 August 2026).
teacherTeachers must decide whether to turn Gemini on in their class and under what rules.school leaderInstitutions using Classroom need to set AI feature availability at admin level.parentYounger children can also access the AI feature through their school account, so it is worth asking about the school's rules.studentThe starter prompts may help, but solving the task remains the student's own responsibility.tool · United StatesKhan Academy founder says AI tutoring works with a teacher's help
Sal Khan says AI teaches well only with teacher support; full essays are not graded by AI, but rubric-based assessment could be possible within a year (CSMonitor, 7 August 2026).
teacherAI feedback can reduce workload, but responsibility for assessment stays with the teacher.edtech companyRubric-based AI assessment may be the next development direction.