Q: What are the essential first steps in how to create K-12 AI policy that protects learning? A: Start by naming the policy’s purpose—protecting evidence of learning, student well-being, and academic integrity—and commit to equity, privacy, and responsible use. Define common terms in plain language and set clear practices like levels of allowed AI use, disclosure requirements, assessment redesign, detector cautions, privacy protections, staff training, and regular review. Q: How should schools set levels of allowed AI help for assignments? A: Schools should define levels such as Level 0 (no AI allowed), Level 1 (planning only), Level 2 (support with disclosure), and Level 3 (AI as a tool partner for tasks like coding or translation). Each assignment should be matched to a level with an explanation of why that level is appropriate for the learning outcome. Q: What should student disclosure of AI use include? A: Students should note whether they used AI, explain for what purpose, and record the prompts they used when possible. Attaching AI chat logs or screenshots is recommended to make the process transparent and help teachers assess evidence of student thinking. Q: Can schools rely on AI-detection tools as definitive proof of cheating? A: No, detectors can make mistakes in both directions and should not be the sole basis for punishment. One study cited false-positive rates as high as 50% and false-negative rates as high as 100% depending on the tool, with nonnative English writing falsely flagged at about 61.3% and other studies showing substantial misclassification of AI-generated text after editing. Q: How can teachers redesign assessments to ensure work shows student learning? A: Collect process artifacts like notes, outlines, drafts, and comments, add short oral checks or in-class quick writes, and use paper-and-pencil tasks for key moments to verify independent skill. Grade process and product together and ask for reflections that explain choices to make student thinking visible. Q: What privacy protections should districts require when approving AI tools? A: Approve only tools that meet student data laws and explicitly prohibit training models on student work, and set chat history or data sharing off by default. Also provide non-AI options for required tasks so students who opt out or have privacy concerns are accommodated. Q: What fair integrity procedures are recommended when work appears suspicious? A: Use a “conversation first” approach, inviting students to explain their process and allowing revisions rather than relying solely on a detector score. Apply consistent consequences focused on learning recovery and use interviews and process evidence to determine understanding. Q: How should leaders implement and iterate an AI policy after launch? A: Launch with a cross-role team, map assignments to allowed-use levels, approve a short list of safe tools and settings, share a one-page student guide, host a parent night, and train staff on assessment redesign and incident response. After launch, collect sample work with process artifacts, review detector use for bias, survey students and teachers, track signals like fewer integrity cases and more process evidence, and refine policy elements each term.