What's real in AI for your classrooms this week, and what's just loud. A weekly digest of the most practical findings and headlines for the K-12 education leader.
On August 7, EdWeek published the contents of the STUDENTS FIRST Act, the nine-page framework 98 high school students from all 50 states drafted and voted into being, 82 to 16, at America's Youth AI Festival in Boston in July. This newsletter banked that event in July as an unpublished WATCH, because organizers would not let reporters read the document. The text is now public, and AASA is sharing it with its 10,000-plus member network. The provisions are concrete: AI cannot be the sole determinant of a grade, every syllabus needs a written AI policy per assignment, teachers can require an oral defense when AI use is suspected, and a student can refuse to use AI on an assignment and get an alternative instead. A Florida science teacher told EdWeek he already added the oral-defense option to his syllabus this fall. The students are honest about what they left out too: nobody worked out who pays for the laptops and training their own recommendations require. Education Week →
EdWeek reported August 6 on how schools are updating cyberbullying policy to name AI-generated deepfakes explicitly, following an Illinois law requiring every school in the state to do so. The concrete moves, per two Illinois principals and a Center for Democracy and Technology policy counsel: name deepfakes in the policy rather than assume they're implicitly covered, state plainly that creating and sharing nonconsensual images is a criminal offense and not just a school infraction, designate one point person for reports, and pair discipline with a restorative process, a "reentry meeting" before a suspended student returns to class. The federal Take It Down Act requires platforms to remove nonconsensual imagery within 48 hours of a verified request, but says nothing about what a school does the week before that happens. The throughline from CDT's Kristin Woelfel: "here is who you report this to" has to be written down, not assumed. Education Week →
Legend.org, a for-profit company, bought the nonprofit Mastery Transcript Consortium last month and is building AI tools for its 400 member schools to grade beyond A-F: papers, exams, recorded presentations, even artwork and music. Co-founder Matt Sornson describes video-based AI evaluation of "tone of voice... body language and posture... eye contact" for communication skills. Every technical claim in the reporting comes from Sornson or consortium founder Scott Looney, both of whom have a stake in the sale landing well, and no independent test of accuracy against these subjective rubrics is cited. Sornson says the AI produces a first-pass score a teacher must finalize, not a verdict, though Looney separately predicts schools "will be surprised" if they resist tools he calls "highly accurate." Nobody has published what "highly accurate" means for scoring a kid's eye contact during a speech. The 74 →
Every story this week is a specific question that got an answer, or didn't. Varsity Tutors could not name the researchers behind its one success story, and its own funder found only 15 percent of students hitting their recommended tutoring dosage, so the district money followed the answer, not the pitch. A hundred students spent three days writing down exactly what they want asked before a class uses AI: can I refuse, is my grade final without a human, will the rules be in the syllabus. Legend.org has an answer for what its AI grades and no answer yet for what "highly accurate" means when the rubric is a kid's eye contact. And two Illinois principals show what it looks like to write the deepfake question down before you need it: who do I report to, and what happens next. The pattern under all four: the schools and vendors that held up this week are the ones with a written answer ready, not just a confident one.
Your one move this weekPull up your cyberbullying policy and search it for the word "deepfake." If it isn't there, add one paragraph: the definition, the criminal exposure, and the name of the person a student reports to. It's the cheapest fix in this issue, and per this week's other three stories, cheap and specific is the version that holds up.
Quality AI lesson planning that lifts instructional quality and cuts teacher workload at the same time. It closes the gap between your curriculum and student-ready instruction, and gives leaders visibility into how your instructional model is actually being implemented, without piling more onto teachers.
See Planning Partner →