AI & Society Brief — September 2026

This month institutions stopped debating artificial intelligence and started writing rules about who may use it, for what, and who answers for the result — while the measured effects on employment turned out to be more uneven than the announcements suggested.

New York bars generative AI for 600,000 younger students for a full school year

• Source: NYC Mayor's Office / NYC Public Schools · 2 September 2026 · https://www.nyc.gov/mayors-office/news/2026/09/mayor-mamdani-and-chancellor-samuels-put-students-first-with-nat — corroborated by Chalkbeat, 2 September 2026 · https://www.chalkbeat.org/newyork/2026/09/02/nyc-schools-to-set-ai-policy-ban-screen-time-limits/
• New York City Public Schools announced on 2 September 2026 a one-year moratorium on student-facing generative AI and companion chatbots from pre-K (2-K) through grade 8, effective for the 2026-27 school year. It covers roughly 600,000 students, about two-thirds of the system, and withdraws around 40 software tools; exceptions cover assistive technology for students with disabilities, multilingual learners, and computer-science and career-readiness programmes. The same policy recommends daily screen-time caps of 30 minutes in grades 3-5 and 45 minutes in grades 6-8, opens pilots for up to 50,000 high-school students, and allows teachers to use approved AI for lesson planning and administrative drafting but not for grading, behaviour monitoring, counselling or special-education plans.
• Why it matters: The largest school district in the United States separated two questions most organizations still treat as one — whether adults may use a tool, and whether the people they are responsible for may. Any institution that deployed AI tools by default now has a documented precedent for withdrawing them without declaring the technology harmful.

Chinese courts rule that being replaced by AI is not a lawful reason to dismiss someone

• Source: Yahoo Finance (syndication) · 14 August 2026 · https://finance.yahoo.com/technology/ai/articles/chinas-courts-rule-workers-cant-153000752.html — second account: Moneywise · https://moneywise.com/news/top-stories/china-courts-workers-ai-displacement-labor-laws
• The Guangzhou Intermediate People's Court upheld a lower-court ruling that a company had unlawfully dismissed a graphic designer after AI took over his work, with judge Chen Shiyuan stating that employers cannot transfer the risks of ordinary technological change onto employees. In a separate case, a Hangzhou district court ruled in August 2025 against a technology firm that dismissed an employee, surnamed Zhou, who checked large-language-model output and had refused reassignment at a 40% pay cut (from 25,000 to 15,000 yuan a month); the firm lost its appeal in April 2026. Under China's Labour Contract Law, dismissal requires proof of a "major change in objective circumstances", and the courts held that replacement by AI alone does not meet that test where redeployment was possible.
• Why it matters: In a jurisdiction where automation is advancing fast, courts placed the cost of a technology decision on the employer, not the employee. Organizations that justify a restructuring by AI adoption should expect the legal question to be redeployment, not capability.

Kenya's essay-writing economy shrank, and its writers' incomes with it

• Source: The New York Times · 5 September 2026 · reprise Business Standard, 6 September 2026 · https://www.business-standard.com/world-news/kenyans-did-college-students-homework-for-years-then-ai-arrived-126090600083_1.html — secondary account: Gizmodo, 7 September 2026 · https://gizmodo.com/spare-a-thought-for-the-kenyan-essay-writers-who-used-to-help-students-cheat-before-ai-2000808074
• At its peak early this decade, at least 40,000 people in Nairobi were paid to write coursework for students abroad; the reporting documents a collapse of that market within roughly two years of generative chatbots becoming widely available. Writers who remain in the trade have seen monthly earnings fall from about $900-1,200 to about $500-800. One writer, Teresios Bundi, produced more than 2,500 papers over twelve years at $40-70 each — at least five times what local public-health work paid him — and has since left the business; transcription and translation gig work in Kenya was disrupted along similar lines earlier.
• Why it matters: This is one of the first documented cases of an export service sector in a lower-income country losing its market to AI rather than to another country's labour costs. Organizations whose revenue depends on remote, text-based service delivery can read here what a two-year timeline of substitution actually looks like in earnings, not in forecasts.

In August, AI dropped to fourth among the reasons US employers gave for cutting jobs

• Source: Challenger, Gray & Christmas · report published 2 September 2026 · https://www.challengergray.com/blog/challenger-report-august-job-cuts-up-58-consumer-products-food-lead/
• US employers announced 52,881 job cuts in August 2026, down 38% year on year and the lowest August total since 2022. Artificial intelligence was cited in 3,462 of those cuts — the fourth-most-cited reason and the lowest monthly figure attributed to AI since December 2025, ending a five-month run in which AI had led every month; restructuring led August with 16,173. Year to date, AI has been cited in 116,175 announced cuts, about 22% of the total and still the leading annual reason, while announced hiring from January to August reached 119,825, up 37% on the same period in 2025.
• Why it matters: The figures track what employers say publicly about their own decisions, not an independent audit of causes — which is precisely why a sharp monthly swing is worth noting: the reason given for a job cut is itself a communication choice. Anyone planning headcount on the assumption of a steady AI-driven contraction is working from a trend that did not hold in August.

Most American teens use AI for schoolwork; few have ever discussed it with a teacher

• Source: Common Sense Media, Youth AI Safety Institute, « Teens in the AI Era: Schoolwork and Skills That Matter » · 18 August 2026 · https://www.commonsensemedia.org/research/teens-in-the-ai-era-schoolwork-and-skills-that-matter
• A nationally representative survey of 1,017 US teenagers aged 13-17 found that 70% use AI for schoolwork. Among those users, 63% obtain an answer from the tool: 25% submit it as is, 31% rewrite it in their own voice, and 35% improve it with what they already know. Only 30% say a teacher has ever discussed with them how to use these tools safely. The study was published by Common Sense Media, a US non-profit that advocates for children's online safety and therefore has an interest in the salience of the issue.
• Why it matters: The gap is not between use and non-use, but between use and supervision — a majority of users are operating without any instruction from the adults who assess them. For anyone running a training programme, the practical implication is that the rules of use have to be taught explicitly, because they are not being inferred.