Sources & Verification

Sources consulted in developing and reviewing The AI House Rules Kit

Reviewed July 2026. Seller-maintained:

This is a targeted verification list, not a systematic review or complete bibliography. We prioritized current U.S. government or first-party sources for response routes and original peer-reviewed research where practical for selected empirical claims. AI House Rules rechecks this source set before each published update. The family-conversation frameworks are practical editorial guidance, not tested clinical protocols. This edition has not been independently reviewed by a licensed clinician or attorney. This page is not legal, medical, or mental-health advice.

Retrieval practice

Karpicke and Blunt's controlled study found that practicing retrieval produced stronger meaningful learning than elaborative concept mapping for the materials and tests studied. The Kit uses retrieval as a study tool—not as a claim that one method fits every learner or task.

Karpicke & Blunt (Science, 2011) — PubMed record (opens in a new tab)

AI-detector false positives

Liang and colleagues found that widely used GPT detectors frequently misclassified human writing by non-native English writers. Turnitin’s current guidance says its AI Writing Report should not be used as the sole basis for adverse action against a student. A detector score is therefore a signal to investigate, not proof of misconduct.

Asking directly about suicide

NIMH states that asking directly about suicidal thoughts does not cause or increase them. Ask calmly, listen without judgment, and treat every disclosure seriously.

988 crisis support

In the U.S. and its territories, 988 provides free, confidential crisis support by call, text, or chat, 24/7/365. Call 911 for an immediate life-threatening emergency.

Child exploitation, sextortion, and image removal

NCMEC's CyberTipline is the centralized U.S. reporting system for suspected online child sexual exploitation. NCMEC advises getting help before paying or complying with sextortion demands and preserving messages without creating extra copies of sexual imagery involving a minor. Take It Down can hash eligible files on-device to help participating platforms limit spread; it does not guarantee removal everywhere.

Poison Help

The HRSA-funded Poison Help line connects callers in the U.S. and many territories to a local poison center at **1-800-222-1222**. Call right away for a possible poisoning; do not wait for symptoms.

HRSA — Poison Help (opens in a new tab)

Platform age and account rules

Age, consent, region, and supervised-account rules differ by service and feature. Use accurate birth dates and only an arrangement the service expressly permits. OpenAI currently requires a minimum age of 13 (or the local age of consent) and parent or guardian permission under 18; Google documents a parent-enabled supervised Gemini experience for eligible children under 13 in supported regions.

U.S. schools, accommodations, and Title IX

U.S. Department of Education guidance treats assistive technology and disability-related modifications as individualized supports; follow the student's written IEP, Section 504 plan, or school-approved accommodation. Separate guidance identifies nonconsensual sharing of real, altered, or AI-created intimate imagery as a form digital sexual harassment can take. Whether Title IX requires a particular response depends on the facts and the school's education program or activity.