K-12Issue 245 records
K–12
Fresh K–12 signal on 29 Sep centers on districts writing enforceable AI operating rules while scrutiny sharpens on emotional/companion use and vanity metrics. Same-day Petaluma reporting shows a California draft policy with grade-banded access (no student-facing AI in K–2; literacy-only in 3–5; closed or teacher-guided secondary use), teacher-owned grading, privacy-vetted tools, and forceful parent warnings about cognitive offloading and chatbot mental-health harms. Quaker Valley’s board AI policy supplies a concrete FERPA-gated approved-tool process and a 0–4 assignment AI-use scale with human-final grading and HR decisions. Against that adoption wave, edCircuit’s 28 Sep measurement brief argues logins and ‘hours saved’ do not prove students learn more and urges unaided independence checks and stop criteria—echoing NCES’s gap between widespread teacher AI use and low leader confidence in better outcomes. On the risk side, Xiao et al.’s interviews with 33 U.S. teachers introduce AI companion literacy because intimate AI relationships remove developmental friction, and EdTech Magazine highlights that roughly one in five youth have sought mental-health advice from chatbots, often secretly, without HIPAA-grade protections. Net for SLED: ship age/assignment controls and approved catalogs; treat companion modes as a student-services problem; and refuse to scale on usage dashboards without durable-learning evidence.
- Evidence records
- 5
- Cross-source patterns
- 3
- Evidence classes
- 3 independent reporting1 academic research1 standards or public-body guidance
- Outcomes
- 3 cautionary2 emerging
- Source freshness
- 4 new this fortnight1 recent
- Research completed
- 2026-09-30
Choose a role to see its takeaway beside every record in the ledger.
Patterns across the evidence
Synthesis · Lighthouse Advisory interpretation
Districts are operationalizing AI via grade bands and assignment-level use scales—not open chatbot defaults
Petaluma’s draft continuum (no K–2 student-facing AI; closed middle-school environments; teacher-guided secondary use with disclosure) and Quaker Valley’s 0–4 assignment AI-use scale with FERPA-gated approvals show the emerging operating pattern: constrain by age/assignment, keep teachers as graders, and inventory approved tools before scale.
Operating questionCan the district show a published grade-band or assignment AI-use matrix, an approved-tool list with privacy review, and teacher-final grading—before enabling broad student generative AI?
Supporting evidencePetaluma Argus-Courier / Petaluma NewsQuaker Valley School District
Emotional/companion AI use is a distinct K–12 risk requiring literacy, vetting, and human escalation
Teacher interview research proposing AI companion literacy, EdTech Magazine’s synthesis of youth mental-health chatbot use, and Petaluma parent testimony about dangerous chatbot uses converge: academic AI policies miss secret emotional reliance, simulated empathy without clinical judgment, and the need for counselor/parent shared responsibility beyond device bans.
Operating questionDoes AI guidance explicitly cover emotional/companion use, disable school companion modes, and give students a clear stop-and-seek-human path?
Supporting evidenceCarnegie Mellon University / multi-institution HCI education research (arXiv:2609.13479)EdTech MagazinePetaluma Argus-Courier / Petaluma News
AI success metrics must prove durable learning and net workload—not activations or assisted polish
EdCircuit’s measurement framework insists on unaided demonstrations, equity cuts, and stop criteria, while Petaluma’s draft and parent debate show adoption pressure without outcome evidence. Together they reinforce that assisted completion and literacy frameworks are not substitutes for retained learning checks.
Operating questionBefore renewing any student-facing AI tool, can leaders show independent performance without the tool and a written expand/modify/stop rule?
Supporting evidenceedCircuitPetaluma Argus-Courier / Petaluma News
Evidence ledger
Full record · every source keeps its link and limitations
Petaluma City Schools draft AI policy: grade-banded access, teacher grading, and parent developmental-risk pushback
Amelia Richardson reports that Petaluma City Schools’ Technology and Innovation Committee presented a still-unadopted draft AI board policy at the Sept. 15 board meeting: staff may permit ethical AI as a supplemental classroom tool only when systems are district-vetted; teachers retain all grading/assessment; K–2 would have no classroom AI use, grades 3–5 a basic introduction with no student-facing tools, grades 6–8 responsible use in closed school-focused environments, and grades 9–12 guided use (e.g., Canva, possibly Gemini) at teacher discretion with disclosure required. Parents and a trustee raised developmental, cognitive-offloading, and mental-health/safety risks—including documented chatbot-linked harm cases—and urged prioritizing books and human interaction; no board vote was scheduled.
Why it matters, evidence and limitations
- Why it matters
- Same-day California district signal showing grade-banded AI access, privacy-vetted tools, and public companion/developmental-risk pushback during board policy drafting.
- Evidence and measured results
- Local newspaper account of board presentation and parent comment; draft not yet ratified; no outcome metrics.
- Limitations and uncertainty
- Draft policy not adopted; single-district journalism; parent/trustee claims about global score declines are not independently verified here.
edCircuit: Measure AI by durable learning and independence—not logins, prompts, or claimed time savings
edCircuit’s Sept. 28 analysis argues districts commonly report AI success via account activations, prompt counts, and estimated teacher time saved while failing to answer whether students learn more. Citing NCES 2025 (73% of public schools report some teacher AI use vs only 38% of leaders agreeing AI will improve outcomes) and REL Central’s note that research has not established which teacher AI uses improve learning, it urges pre-defined goals, net workload (time saved minus verification), unaided/independent student demonstrations after tutoring support is removed, equity subgroup review, and advance criteria for expand/modify/stop—including willingness to end ineffective pilots.
Why it matters, evidence and limitations
- Why it matters
- Operational measurement checklist that converts the assisted-vs-retained-learning brief into board-ready evaluation practice for AI pilots.
- Evidence and measured results
- Practice synthesis citing NCES/REL/NIST; not a new primary RCT.
- Limitations and uncertainty
- Vendor-neutral commentary; NCES/REL citations secondary; does not present new causal student-outcome estimates.
arXiv study: 33 U.S. teachers call for AI companion literacy beyond tool-focused AI literacy
Xiao and colleagues interviewed 33 U.S. K–12 teachers with scenario cards spanning AI as study buddy, tutor, friend, romantic partner, and parent-like figure. Teachers welcomed academic companions but worried intimate companions remove developmental friction needed to sustain human relationships; 24 of 33 named romantic-partner scenarios most concerning. Intervention followed jurisdiction (school visibility, academic vs relational role, safety override) more than perceived harm alone. Authors propose AI companion literacy—recognizing relational engagement, disclosure limits, and dependence—as shared work across teachers, counselors, parents, platforms, and policymakers, noting bans cover school devices/hours while much companion use is on personal phones.
Why it matters, evidence and limitations
- Why it matters
- Newly relevant alongside same-day Petaluma parent safety concerns and prior companion bans: supplies teacher-grounded curriculum/governance framing for emotional AI use.
- Evidence and measured results
- Qualitative scenario-based interview study (n=33); not an outcome RCT; Prolific-recruited sample limits representativeness.
- Limitations and uncertainty
- Hypothetical scenarios may differ from real cases; U.S.-only sample; published Sept 11—why-now is pairing with 29 Sep district/parent companion-risk discourse and thin prior archive coverage of companion literacy.
Quaker Valley AI policy: FERPA-gated tools, assignment AI-use scale, and human-final grading/HR decisions
Quaker Valley’s Artificial Intelligence policy PDF requires age-appropriate AI instruction emphasizing human supervision and skepticism; CIPA filtering of AI tools capable of generating obscene/pornographic material; strict FERPA adherence; and a ban on student PII in AI tools not pre-approved by the Director of Technology after privacy/data-handling/FERPA review. Teachers must state permitted AI use per assignment on a 0–4 scale (No AI through Full AI Use with Human Oversight) with disclosure rules; AI may assist grading but teachers make final grade decisions; final HR evaluation/hiring/promotion decisions remain human; and regular PD on ethical AI, privacy, and compliance is required.
Why it matters, evidence and limitations
- Why it matters
- Primary LEA policy artifact operationalizing FERPA procurement gates and assignment-level AI-use scales—useful template alongside same-week Petaluma draft debates.
- Evidence and measured results
- District policy PDF retrieved in full; local press reports 8–0 board approval Sept. 22; not outcome evidence.
- Limitations and uncertainty
- Single Pennsylvania district; PDF shows template numbering ('No. XXX'); effectiveness depends on classroom enforcement not examined here; why-now is fresh board adoption plus parallel Petaluma grade-band drafting.
EdTech Magazine: Student AI chatbot use for mental health advice is common, often secret, and not a care substitute
Rachelle Dené Poth’s Sept. 24 EdTech Magazine analysis argues academic AI misuse is only part of the risk: a 2026 JAMA Pediatrics survey found about 1 in 5 U.S. youth ages 12–21 used an AI chatbot for mental health advice, nearly two-thirds telling no one; Common Sense reports 72% of teens tried an AI companion. Practitioners warn chatbots simulate empathy without clinical judgment or HIPAA coverage, may retain/use sensitive disclosures, and should not replace counselors. Recommended district responses include anonymous student listening, emotional-use boundaries in AI guidance, wellness-product vetting (age, evidence, crisis escalation), and stronger human support networks.
Why it matters, evidence and limitations
- Why it matters
- Independent scrutiny piece connecting teen mental-health chatbot use to district policy gaps—pairs with Petaluma parent testimony and the companion-literacy study.
- Evidence and measured results
- Practitioner analysis citing JAMA Pediatrics/Common Sense/Pew/APA; secondary citation of survey figures; not a new district audit.
- Limitations and uncertainty
- Survey statistics are cited second-hand; author is also a consultant; Sept 24 predates last k12 run—why-now is direct relevance to 29 Sep Petaluma safety debate and unarchived companion-literacy framing.
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.