K-12Issue 195 records
K–12
One day after archived coverage of NYC/LA generative-AI reverberations, fresh K-12 signal shifts to capacity, training, and dual operating paths. A 23 Sep 2026 Educator K/12 report on an American College of Education survey of 694 U.S. teachers finds 54% still have no AI training even as nearly half expect AI grading/planning and many anticipate more tutoring—amid overtime, burnout, and doubts that AI will make teachers more valuable. Campos and Singleton’s national principal survey (June 2026 PDF) shows the deeper structure behind that gap: AI use spread far faster than written policies (58% none) and tutoring contracts (76% none), with lower integration in disadvantaged, charter, and private schools; district scale—not principal enthusiasm alone—closes about a third of the public disadvantage gap. Seattle Times (18 Sep) adds a developmental/privacy control: Washington’s new companion-chatbot rules (hourly AI disclosures for minors; self-harm protocols; Jan 1 effective) while Seattle Public Schools enables vetted tools under human-centered guidance. Pharr-San Juan-Alamo ISD’s September-updated AI Guidebook illustrates an affirmative path—approved tool lists, FERPA bans on identifiable student prompts, and student pilot/co-pilot integrity rules—contrasting megadistrict pauses. An August arXiv detector study warns that false-positive rates on human academic writing can span 0–100% and that polished editing style confounds scores, limiting detector-only misconduct decisions for multilingual learners. Net: SLED districts should fund policy, PD, and age-appropriate companion safeguards as first-class work—and treat enablement vs pause as a capacity choice, not a feature race.
- Evidence records
- 5
- Cross-source patterns
- 3
- Evidence classes
- 2 independent reporting2 academic research1 standards or public-body guidance
- Outcomes
- 3 cautionary2 emerging
- Source freshness
- 2 new this fortnight2 recent1 older, newly relevant
- Research completed
- 2026-09-25
Choose a role to see its takeaway beside every record in the ledger.
Patterns across the evidence
Synthesis · Lighthouse Advisory interpretation
Teacher/tool adoption still outruns training and written policy
Fresh ACE/Educator teacher survey evidence (54% untrained) and Campos/Singleton principal survey evidence (58% no written GenAI policy; shallow integration) jointly show U.S. K-12 adoption remaining ahead of institutional scaffolding—raising workforce and equity risk if licenses scale first.
Operating questionBefore expanding GenAI licenses, what share of teachers are trained and what written policy is live—especially in higher-need schools?
Supporting evidenceThe Educator K/12 reporting American College of Education surveyChristopher Campos (University of Chicago) and John Singleton (University of Rochester)
Enablement and pause both need explicit youth-safety and privacy operating rules
Seattle Times documents WA companion-chatbot disclosures/crisis protocols alongside SPS enablement, while PSJA’s September guidebook shows an affirmative approved-list path with hard FERPA prompt bans—together arguing that either strategic path still requires concrete safety/privacy mechanics, not slogans.
Operating questionFor each AI product, is it instructional or companion, what youth disclosures apply, and is identifiable student data banned from prompts?
Supporting evidenceThe Seattle TimesPharr-San Juan-Alamo Independent School District (Texas)
Integrity enforcement cannot rest on detector scores alone
The arXiv paired-edit study’s extreme FPR range and style confounds, read alongside districts tightening student AI rules (PSJA disclosure/permission norms; WA/NYC companion controls in Seattle Times), caution that automated AI-writing scores are unreliable sole evidence—especially for multilingual learners.
Operating questionIf a detector flags student work, what human process evidence is required before any academic sanction?
Supporting evidenceChung-Ang University and Wordvice (Park, Jeong, Kim)Pharr-San Juan-Alamo Independent School District (Texas)The Seattle Times
Evidence ledger
Full record · every source keeps its link and limitations
ACE teacher survey via Educator Online: 54% lack AI training amid rising expectations and burnout
On 23 Sep 2026, The Educator reported an American College of Education survey of 694 U.S. teachers finding 54% received no training on using or managing AI at school, while 49% expect AI-assisted grading, planning, and administration to grow over 5 to 10 years and 43% expect more AI tutoring. Among teachers naming AI a top classroom challenge, 44% still report no training. Only 22% believe AI will make teachers more valuable. Context includes 7.7 average weekly overtime hours, 36% high/very high burnout, and 44% considering leaving within five years. ACE Interim CEO Feikema argues system leaders own the training gap and must make room for AI literacy by replacing other PD or reducing low-value admin—not stacking new training atop overload.
Why it matters, evidence and limitations
- Why it matters
- Directly informs U.S. K-12 districts evaluating GenAI rollouts: procurement without funded, job-embedded PD and clear assessment/privacy expectations risks teacher scepticism and burnout amplification. Distinguishes assisted teacher productivity hopes from unverified student-learning gains.
- Evidence and measured results
- Secondary journalism quoting ACE survey statistics and an on-record interview with ACE Interim CEO/Provost Dr Jodi Feikema; article does not publish full sampling frame, response rate, or questionnaire.
- Limitations and uncertainty
- ACE has institutional interest in educator credentialing; survey microdata not opened here; overtime/burnout figures may come from linked ACE research not fully detailed in the article.
National principal survey: AI use outruns policy/training, with disadvantage and sector integration gaps
Campos and Singleton’s June 2026 working paper reports an original national survey of K-12 principals (fielded Dec 2025 to Feb 2026) measuring institutional AI integration. Teacher AI use spread rapidly while scaffolding lagged: 58% of principals report no written generative-AI policy; 76% report no AI-tutoring license/contract; 67% did not use ESSER funds for AI. An integration index combining policy, training, student-use guidance, leadership, and tools shows schools serving more disadvantaged students score 0.07 to 0.11 SD lower per SD of disadvantage, and charter/private schools score 0.23 to 0.44 SD lower than traditional public schools. District size accounts for roughly one-third of the public-school disadvantage gap; principal enthusiasm and budgets raise integration levels but do little to close gaps. Why now: after 2026-09-23 coverage of NYC/LA pauses reverberating nationwide, this evidence reframes pause-versus-enablement as a system-capacity equity problem, not merely a values debate.
Why it matters, evidence and limitations
- Why it matters
- Gives SLED buyers an equity lens for GenAI: buying tools without district-scale implementation capacity can widen integration gaps for disadvantaged and non-district schools. Supports engagements that build shared policy, PD, and vendor infrastructure rather than school-by-school experimentation.
- Evidence and measured results
- Original survey with school-tied invitations merged to federal demographic data and reweighted; randomized participation incentives used to probe selection; correlational gaps with controls; not an achievement RCT.
- Limitations and uncertainty
- Principal self-report; integration is not learning outcomes; June 2026 vintage requires why-now framing; charter/private lag may partly reflect different governance models.
Seattle Times: Washington companion-chatbot youth safeguards contrast SPS enablement with NYC/LA pauses
Sep 18 Seattle Times reports Washington’s spring 2026 law requiring companion chatbots to notify minors they are interacting with AI at the start and at least once per hour; use reasonable measures against manipulative engagement and sexual content with minors; disclose non-healthcare-provider status for health advice; and create self-harm/suicidality detection with crisis referrals. Rules take effect Jan 1 and are enforceable via the Consumer Protection Act. The piece contrasts Seattle Public Schools’ ethical-AI enablement (tools such as MagicSchool AI, Canva, Adobe Firefly) and OSPI human-centered guidance with NYC’s yearlong GenAI limits for elementary/middle grades plus all-grade companion bans and LAUSD’s temporary GenAI block on student devices. Also notes OpenAI ChatGPT for Teens protections and Microsoft-AFT school AI safety/privacy standard. Why now: complements 2026-09-23 archived megadistrict-pause coverage with a state-level companion-safety control path.
Why it matters, evidence and limitations
- Why it matters
- Gives SLED districts a concrete companion-chatbot control checklist and shows enablement and pause can coexist with youth mental-health safeguards. Relevant to developmental-risk evaluation in the K-12 brief.
- Evidence and measured results
- Journalistic synthesis of enacted WA requirements, district website statements, and national comparisons; not a statute annotated code extract or outcome evaluation.
- Limitations and uncertainty
- Secondary reporting; Ballmer Group funds Mental Health Project with stated editorial control retained; SPS tool inventory may change; effectiveness of chatbot crisis detection unmeasured here.
PSJA ISD September-updated AI Guidebook: approved tools, FERPA prompt bans, student pilot/co-pilot rules
PSJA ISD’s AI Guidebook, marked Updated September 2026, sets an affirmative GenAI operating model: staff must use district-approved tools via district login and must not enter, upload, paste, or describe identifiable student information into public AI tools (FERPA/Board Policy FL), with incidents reported to IT helpdesk. Approved tools listed include ChatGPT, Microsoft Copilot, Gemini, MagicSchool AI, NotebookLM, Wayground, Khanmigo, Canva AI, Adobe Express, Diffit, Brisk, Curipod, and TeachShare; new tools require IT approval. Student expectations emphasize permission, citation, fact-checking, no private info, and no copy-paste misuse under a pilot / AI co-pilot framing, with academic integrity requiring teacher permission and disclosure. References Texas DIR AI code of ethics obligations (Gov Code 2054.702 / TAC 219). Why now: provides a concrete enablement template contrasting NYC/LA pause pathways covered 2026-09-23.
Why it matters, evidence and limitations
- Why it matters
- Useful SLED artifact for districts choosing guided enablement: shows how approved lists, FERPA prompt bans, and student integrity rules can be packaged together. Not evidence of learning gains.
- Evidence and measured results
- Primary district policy webpage inspected in full HTML text extraction; living document without independent outcome metrics.
- Limitations and uncertainty
- Single-district guidance; enforcement and training quality unknown; approved list includes general-purpose tools that still need classroom controls; not an evaluation of learning or safety outcomes.
Large paired-edit study: AI detector false positives span 0 to 100% and track editing style
Park, Jeong, and Kim analyze 135,389 original vs professionally edited non-native academic manuscript pairs across 13 AI text detectors. False-positive rates on human-written texts ranged from 0.0% to 100.0%. The same edits increased AI scores in some detectors and decreased them in others, with score changes correlating with editing intensity; practical significance thresholds were met in only 7 of 13 detectors. Authors identify professional editing style as a key confound rather than cleanly separating origin from linguistic form, raising fairness and reliability concerns in academic settings. Why now: as districts tighten integrity rules alongside AI pauses/pilots, detector-only misconduct findings are especially risky for multilingual K-12 writers.
Why it matters, evidence and limitations
- Why it matters
- Supports K-12 academic-integrity policy that forbids sole reliance on AI detectors—aligned with developmental fairness for multilingual students and with process/evidence-based misconduct review.
- Evidence and measured results
- Large-scale paired natural experiment holding content/authorship fixed while varying professional editing; detector outputs compared pre/post; academic manuscripts 2018 to 2025, not K-12 classroom essays.
- Limitations and uncertainty
- Corpus is professional academic editing, not K-12 essays; one coauthor affiliated with editing vendor Wordvice; detector landscape changes quickly; transfer to schools requires caution.
How to read this edition
Source findings, measured results and limitations come from the cited publications. Patterns, operating questions, role takeaways and implementation considerations are Lighthouse Advisory interpretation, stated as questions to validate locally rather than guaranteed outcomes. Vendor and operator claims are labeled as claims. Full research method.
- Independent reporting
- Independent reporting with attributable sources but without a formal evaluation design.
- Academic research
- Research produced through an academic institution or peer-reviewed venue.
- Standards or public-body guidance
- Normative or advisory guidance from a standards body or public institution.