Three friendly AI robots rowing together across glowing digital waves

The Robots Are Making Waves

iTeachAI NEWS

Edition 27 | July 11, 2026

🤖 Play the Robot Row easter egg

Row, row, row your bots: the AI fleet left the harbor this week

This week the starting gun fired. Two weeks ago this newsletter told you Washington was previewing OpenAI's next model behind closed doors. On Thursday the doors opened: GPT-5.6 went public in three tiers named Sol, Terra, and Luna, alongside an agent built to do whole jobs instead of answering questions. The same week, Illinois published the AI guidance every district was waiting on, Ohio's policy deadline came due, the nation's largest school system froze its edtech spending until its own AI rules land, and federal dollars started flowing into summer AI training for teachers. The race is loud, the rules are arriving, and both matter for your classroom. Here is what happened, and what to do about it.

Big Picture

Sol, Terra, and Luna Are Out: The Model Washington Previewed Is Now Yours

On Thursday, July 9, OpenAI released the GPT-5.6 family to the public: Sol, the flagship built for complex reasoning and long-running tasks; Terra, the everyday workhorse; and Luna, the fast, inexpensive one. All three read up to a million tokens of context, roughly a full curriculum binder in a single conversation. Two weeks ago this launch was gated behind a national-security preview; the review concluded and the rollout reached everyone within a day. Pricing for builders runs from one dollar to five dollars per million words read, which is the quiet headline: frontier AI keeps getting cheaper.

Why it matters: The follow-through matters as much as the launch. The preview-then-release pattern worked, so expect it to become the norm for frontier models, and expect the tools your district buys to swap engines under the hood this fall without asking. If your AI guidance names a specific model, rewrite it to name capabilities and rules instead. Models now change faster than policies can.

Read the full story at Engadget →

Classroom Tools

ChatGPT Work: The Agent That Does Whole Jobs, Not Just Answers

Alongside the models, OpenAI shipped ChatGPT Work, an agent designed to carry out entire jobs: gather the inputs, do the steps, deliver the finished product. Think less "draft this email" and more "reconcile this roster against these three spreadsheets and flag the mismatches." It is the clearest signal yet that the industry's bet has moved from chatbots to agents, software that acts rather than suggests.

Why it matters: For schools, agents are where the productivity gains and the risks both live. An agent that assembles your accreditation binder is a gift; an agent with access to student records is a governance question nobody's acceptable-use policy anticipated. Two rules keep you safe ahead of the paperwork: agents get the minimum access needed for the task, and a human signs off before anything an agent did touches a grade, a record, or a family.

Read the full story at VentureBeat →

Industry

Who Is Rowing to Number One? Read the Scoreboard Carefully

OpenAI's launch slides claim Sol beats Anthropic's Claude Fable 5 by 13 points on an agentic-work benchmark. The fine print rows the other way: on SWE-Bench Pro, the hardest widely used software-engineering test, Fable 5 still leads by a wide margin, 80 percent to 64.6, and early independent reviewers report Sol has not surpassed Fable on complex, multi-step work. Meanwhile OpenAI confidentially filed to go public in May, with reports now pointing to a 2027 listing. A blockbuster launch weeks after filing to go public is not a coincidence; it is a pitch deck.

Why it matters: When every lab claims the crown, the benchmark that matters is yours: your lesson-planning prompt, your IEP-summary workflow, your data question. Run the same three prompts through the tools you already have before buying anything new this fall. And enjoy the real dividend of the arms race: capable models keep getting cheaper, which is the best procurement news a district budget has heard in years.

Read the hands-on review at Simon Willison's blog →

Intermission

Speaking of rowing: our tiny AI crew is out on the water, and every click launches another boat. Reach five crews and something nice happens.

🤖 Launch the Robot Row
AI Policy

Illinois Publishes the AI Guidance Every District Was Waiting On

The Illinois State Board of Education released its statewide AI guidance this week, the deliverable its new law required. The framework is notable for what it refuses to do: no mandates, no required tools, no statewide bans. Instead it hands districts model policies and decision tools and keeps two principles at the center: teachers stay in charge of instruction, and student learning, not technology, is the point.

Why it matters: If you teach in Illinois, your district now has its starting document, and board conversations will accelerate this fall. If you teach anywhere else, read it anyway: guidance-not-mandate is emerging as the moderate template states copy from each other, and the questions Illinois answered are the ones your own board will ask. Walking in with another state's answer sheet is a fine way to look prepared.

Read the full story at WSIU →

Big Picture

New York City Freezes Every Edtech Purchase Until Its AI Rules Land

Principals in the nation's largest school system have been ordered to pause all new education-software purchases while the district finishes its AI policy, with updated guidance promised later this summer. Read that sentence again: not AI purchases, all edtech purchases. New York has concluded it cannot tell which tools contain consequential AI anymore, so the whole pipeline waits until the rules exist.

Why it matters: This is the procurement future arriving early. When the biggest buyer in American education says policy comes before purchases, vendors reorganize around it, and smaller districts inherit both the delays and the leverage. If you are eyeing fall purchases, two moves: ask every vendor which model powers their AI features and what happens to student data, and do not sign anything your own one-page AI policy could not defend.

Read the full story at Gothamist →

AI Policy

Ohio's Deadline Just Passed: Every District Now Owes an AI Policy

As of July 1, every traditional public district, community school, and STEM school in Ohio is required to have adopted a formal AI policy, and the state published a model policy so nobody starts from a blank page. Ohio joins a fast-growing club: multiple states now require district AI policies by law, and roughly a hundred AI-in-education bills moved through statehouses this session.

Why it matters: The pattern across Idaho, Ohio, and the states behind them is nearly identical: adopt a policy, protect student data, keep a human in the loop, and do not let AI replace teachers. Which means you can draft one page today that will satisfy whatever your state eventually passes. If your district already missed a deadline, the model policies linked here are the fastest honest path to compliance.

Read the full story at GovTech →

Teacher PD

NSF Puts 11 Million Dollars Behind AI Training for Teachers

The National Science Foundation awarded 11 million dollars to the Computer Science Teachers Association to run AI Professional Development Weeks, intensive summer training designed to reach 2,500 to 3,000 teachers across Indiana, South Carolina, Minnesota, New Jersey, Iowa, Illinois, and more, with a projected downstream reach of some 600,000 students. It is one of the largest federal bets yet that the path to AI-ready classrooms runs through teacher training, not software licenses.

Why it matters: The signal is bigger than the seats: federal money now says teacher PD is the bottleneck in classroom AI, which strengthens every building-level case for funded training time. If you are in a participating state, ask your CSTA chapter about a seat. If you are not, the precedent still works in your favor at budget time: point at it and ask what your district's training plan is.

Read the announcement at NSF →

Security

Deepfakes Are a School-Safety Issue Now, and RAND Wrote the Playbook

RAND's researchers have published a practical report on AI-generated deepfakes in K-12: students using cheap tools to fabricate humiliating or explicit images of classmates and staff, and schools discovering their crisis plans never imagined it. The report treats deepfakes as a school-safety and climate problem, not a technology problem, and lays out response steps: clear reporting channels, victim-first response, preservation of evidence, and teaching students what the law now says about synthetic images.

Why it matters: Every administrator should read this before September, because the first deepfake incident in a building sets the precedent for every one after it. Two things to lift directly: add synthetic media to your bullying and harassment policy by name, and rehearse the first 48 hours the way you rehearse other emergencies. For classrooms, it doubles as the most teachable media-literacy case study there is.

Read the report at RAND →

Try This Week

The theme this week is that the engines changed and the rules arrived, so spend thirty minutes making both work for you. First, pick the one prompt you use most, a lesson-plan request, a rubric, an email to families, and run it through whatever AI you already have, twice: once as you always phrase it, once with one sentence of added context about your students. Save the better output as your new template; that beats any upgrade the labs shipped this week. Second, pull up your district's AI policy, and if there is not one, borrow: Illinois just published model policies and Ohio published a statewide template, both linked above. One page, three answers, done before the school year does it to you: what students may use AI for, what happens to their data, and who reviews AI output before it counts.

A word from iTeachAI Academy
Earn Recert Hours in Your Pajamas 🦝

Every state-aligned iTeachAI Academy course counts toward your recertification, comes with a built-in AI guide that does the heavy lifting, and takes about as long as a good episode of your show. AI integration, structured literacy, MTSS, SEL, and your state's mandated topics, all online, all on your schedule.

One course, $25. All-access, all year, just $149.

See what counts in your state →

Professional development that respects your weekend.

The robots are rowing, the rules are arriving, and somewhere between the two is a classroom that runs a little smoother this fall. That is the boat we are building.

Until next time,

Dr. Janette Camacho

CEO, iTeachAI Academy

P.S. New this summer: iTeachAI Academy Global. Teacher recertification tracks mapped to each country's own standards and licensing rather than generic content, offered in English and the local language. The United Arab Emirates, Qatar, Saudi Arabia, England, Canada, and Brazil are enrolling now, with more on the way. Explore the world catalog at classes.iteachai.co/world.

Free AI courses at classes.iteachai.co

17 free AI tools at iteachai.co/TeacherTools

Know a teacher who needs this? Forward this email.
Subscribe free at iteachaibot.com

Unsubscribe