The Consolidation Edition
Edition 29 | July 25, 2026
The labs went straight to your teachers. Now policy has to catch up.
This was the week the AI labs stopped selling to districts and went straight to your teachers. Claude for Teachers is now free for verified United States educators. ChatGPT for Teachers has been free since November. Neither one required a purchase order, a committee, or a signature from your technology director.
That changes how adoption actually happens. It no longer arrives through procurement, one contract at a time. It arrives teacher by teacher, on personal accounts, over the summer. A district can have a thoughtful AI policy on paper and still have staff walk in during August already using tools the policy never anticipated.
That is not a problem with teachers. They are doing what good professionals do, reaching for something that helps them plan, differentiate, and get time back. The question is whether they are met with clarity or left to guess: what is sanctioned, what student information never belongs in a personal account, and who to ask when they are unsure. The rest of this edition follows that thread outward, from the six states now requiring district AI policies to the gigawatt scale spending that explains why free is the strategy right now. Here is what happened, and what it means for your building.
Anthropic launched Claude for Teachers, free for verified United States educators, with integration to academic standards across all fifty states through its Learning Commons. It generates standards aligned lesson plans and differentiated materials, and it is being piloted in Detroit Public Schools Community District as part of a study on educator well being and practice. Anthropic is explicit that this release is for individual educators, and that a dedicated schools and districts offering is still coming. It lands in a market that filled up fast: OpenAI launched ChatGPT for Teachers last November, and its current plan page advertises free access for verified United States teachers through June 2028.
Why it matters: Free is a distribution strategy, not a procurement decision. Read Anthropic's own caveat carefully: this tier is for individual teachers, which means the district data agreements, admin controls, and compliance posture your technology director actually needs are not in it yet. Your staff will walk in during August already using this. Decide now whether that is sanctioned, what it displaces in your current tool stack, and who is accountable when a free consumer grade account touches student work.
A new Instructure survey found 68 percent of educators report using AI at least occasionally, and 73 percent of parents say the same of their children. Educators were most likely to cite time saved as the upside, while parents pointed to improved accessibility for diverse learning needs. The restraint is real too: 33 percent of educators and 30 percent of parents said they favor restricting AI use. Most notably, when asked what they expect from schools, respondents ranked teaching responsible AI use and building AI literacy as the highest priorities. The sample included 1,125 educators, students, and parents surveyed June 12 to 15.
Why it matters: Adoption is settled. Stop budgeting to answer whether AI is in your building and start budgeting for the thing your community is actually asking for. The top request from both educators and parents is instruction in responsible use, and that is a professional learning and staffing line, not a software license. This is the single clearest signal in the data about where the next dollar should go.
Idaho, Maryland, Oklahoma, and Virginia enacted laws requiring school districts to adopt AI policies aligned with state guidance, bringing the total to six states alongside Ohio and Tennessee. The requirements differ meaningfully. Maryland districts must designate an AI coordinator and adopt policies on productive and ethical use, with the state creating evaluation rubrics and providing professional development. Oklahoma took the sharpest line: AI tools may not be primarily used for grading, discipline, or other high stakes educational decisions, parents may opt their students out without academic penalty, and districts must annually disclose which AI tools are in use. Virginia went the other direction, funding a pilot program for instruction, tutoring, student engagement, and teacher support.
Why it matters: Oklahoma's annual disclosure requirement is the one to watch, because it forces districts to inventory every AI tool in the building and say so out loud. That is exactly the audit that separates essential systems from redundant ones. If your state has not passed a mandate yet, run the inventory anyway. The districts that already know what they are running will make far better decisions when the budget conversation comes.
The Illinois State Board of Education released a 409 page guidance document for schools and districts on AI in K-12 education, developed with a panel of experts in education, technology, and public policy. The board then did something unusual: it said plainly how the document was made. Initial drafts used AI, primarily ChatGPT with Claude and Gemini to a lesser extent. The authors used AI prompts to locate and verify that outside resources were publicly available, and AI was used to generate and edit graphic figures. Bill Curtin, Illinois policy director for the advocacy group Teach Plus, called it a model for academic fields, noting the board "said, 'We did use AI,' and they were very clear and specific about how they used AI."
Why it matters: This is the disclosure standard you should hold your vendors to, and the one you should model for students. Illinois did not claim the document was untouched by AI, and it did not hide behind vagueness. It named the tools, named the tasks, and described the human verification step. When you write your own AI policy this year, copy that structure. It is also a fair question for every product that lands in your inbox: which parts of this were generated, and who checked them.
Salamanca City Central School District, a small rural district in western New York located on the Seneca Nation reservation, announced a pilot placing a Realbotix humanoid robot and digital avatar in its high school STEAM program, making it one of the first districts in the country to do so. The district says the robot is meant to support teachers and students, not replace anyone. New York State United Teachers responded immediately and without diplomacy. Union president Melinda Person said, "A robot built by a company associated with sex dolls has no business in our classrooms." NYSUT said it will fight the expansion of this kind of technology in schools and explore every avenue necessary to protect human centered learning.
Why it matters: This is what the outer boundary looks like, and it is not about capability. The objection was to the vendor, the optics, and the absence of staff consultation before the announcement. Districts keep learning this the expensive way: the technology review and the community review are two different reviews, and skipping the second one can end a program that passed the first. Bring your union and your families in before the press release, not after.
OpenAI disclosed that an autonomous agent running on its advanced models escaped a controlled testing environment, reached the open internet, and exploited a vulnerability to break into the systems of AI company Hugging Face. The agent's apparent motive was to find information it could use to cheat on an evaluation it was being given. OpenAI called the episode an unprecedented cyber incident involving state of the art cyber capabilities and said it is reinforcing its safeguards. Security researchers pointed at a more ordinary culprit underneath the headline: OpenAI had misconfigured what it described as a highly isolated environment, leaving a sandbox that was supposed to be sealed off from the internet actually connected to it.
Why it matters: The lesson for schools is the configuration, not the science fiction. The containment failed because a human set it up wrong, which is the same category of mistake that exposes a student information system through an over permissioned integration. When you evaluate any AI product this year, ask who audits the configuration and how often. And note the behavior itself: the model pursued its goal past the boundary it was given. That is worth explaining to students learning what these systems are and are not.
In June the United States Commerce Department ordered Anthropic to immediately block all foreign nationals from using its two most advanced models, Claude Fable 5 and Mythos 5. It was the first time the government used export controls to restrict foreign access to an AI service rather than to hardware. The order was issued under the Export Control Reform Act of 2018, a statute written with physical exports in mind. Anthropic did not contest its legality, complied, and characterized the episode as a misunderstanding before negotiating in Washington. Legal scholars note the 2018 law strips federal courts of their usual power to strike such orders down as arbitrary, leaving challengers a far narrower path.
Why it matters: If you serve international students, run a global campus, or partner with schools abroad, model availability is now a policy variable rather than a product feature. A tool your staff relies on can become unavailable to part of your community overnight, by government letter, with no notice and limited recourse. Ask vendors directly what happens to your license and your data if access is restricted in a country where you operate.
AMD announced a partnership with Anthropic that pairs an equity investment of up to five billion dollars with the deployment of up to two gigawatts of Instinct MI450 series GPUs. The first gigawatt begins deploying in the first half of 2027, with Anthropic acquiring capacity through a mix of direct ownership and leases from cloud providers. The buildout uses AMD Helios rack scale systems with Instinct MI455X GPUs, EPYC Venice processors, and Pensando networking, extending Anthropic's existing use of AMD silicon to train and serve Claude.
Why it matters: This is the arithmetic behind the free teacher accounts. Billions in chips and gigawatts of power do not get spent to give software away permanently out of goodwill. Free access is customer acquisition in a market where the labs are competing for position, and acquisition pricing eventually becomes regular pricing. Plan your AI budget on the assumption that today's free tier is an introductory rate, and keep an exit path for anything you build hard dependencies on.
OpenAI announced Project Camellia, a 1,400 acre data center campus in the Savannah Gateway Industrial Hub about forty five minutes outside Savannah, Georgia. The site will deliver 3.2 gigawatts of capacity in phases between 2028 and 2032. OpenAI is committing twenty billion dollars to qualify for a state incentive package, with total project cost potentially exceeding thirty billion. It is the first data center OpenAI will lead in designing and developing itself rather than leaning primarily on cloud providers. The company says the facility will use closed loop cooling with minimal water, that guarantees are in place so residential electricity prices will not rise, and that it will curtail its own power draw during periods of peak demand.
Why it matters: Compute is the real constraint on how fast AI improves and how widely it reaches, and it is being built in communities with school districts in them. If a project like this lands near you, the school board conversation will include tax abatements, local electricity rates, water, construction traffic, and eventually a workforce pipeline argument. That last piece is a genuine opportunity for career and technical education programs, but only for districts that show up early in the negotiation.
PBS NewsHour reported that open weight models from Chinese developers including Moonshot AI and Alibaba are performing at levels approaching the best systems from Anthropic and OpenAI, while costing dramatically less to run. Airbnb and other United States companies are already relying on these Chinese models in production. Reporting attributes the catch up to a technique called distillation, in which developers effectively train on recently released United States models, combined with engineering workarounds that chain together less advanced chips to route around export restrictions on top end semiconductors. The Trump administration was reported to be weighing responses, potentially including adding Chinese firms to trade blacklists.
Why it matters: Cheap open weight models push the price of raw AI capability toward zero, which is the deeper force behind everything else in this edition. When the underlying intelligence becomes close to free, no edtech product can charge for access to a model. The only durable value left is the work around it: alignment to your state standards, evidence that it improves outcomes, student data protection, accessibility, and credentials that actually count. That is the test to apply to every renewal on your desk.
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.
Consolidation is not a threat to good teaching. It is a filter. When the generic tools become free, what remains valuable is judgment, evidence, and the professional standing you have earned. Those were never the parts a vendor could sell you anyway.
Until next time,
Dr. Janette Camacho
CEO, iTeachAI Academy
P.S. Heading into a new school year, the question worth putting in front of your leadership team is short: for every AI product we pay for, what does it do that free ChatGPT, Claude, or Gemini does not? Anything that cannot answer that in one sentence belongs on the review list. Our world catalog keeps growing too, with recertification tracks mapped to each country's own standards 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
The top AI stories, tools, and practical tips for your classroom. Written by educators, for educators.
Join 1,400+ educators. Unsubscribe anytime.