Top 10 AI News This Week — November 9–15, 2026
Nov 16, 2026
Three of this week's strongest items came from inside the academy rather than from a vendor newsroom: Fullerton College's own accreditor reset the accreditation calendar and opened a comment window that closes December 4, the CSU faculty senate voted to aim AI at paperwork rather than instruction, and Stanford researchers screened 1,686 papers down to the 52 that actually tested anything. The companies were quieter than usual — Meta's AI blog has not posted since July and xAI's since October — which made the week's real authority institutional. Here's what community college educators should know.
EdTech & Campus AI
ACCJC moves to a ten-year review cycle — and opens a comment window that closes December 4
On November 13, ACCJC President Mac Powell notified chancellors, superintendents, presidents and accreditation liaison officers at every member institution that the Commission approved the transition from an eight-year to a ten-year comprehensive accreditation review cycle, “effective July 1, 2027, subject to the completion of necessary policy revisions and applicable recognition requirements.” The midterm report is not going away — it is getting heavier. ACCJC will “strengthen the Midterm Review process to include direct engagement with an ACCJC Vice President” and an institutional visit, assessing institutional progress, student achievement, fiscal condition and institutional effectiveness. “Annual monitoring and reporting will continue,” along with financial oversight and substantive change review, and the Commission keeps its authority to require additional reporting or special visits at any point.
Three days earlier, on November 10, ACCJC opened six policies for first-read comment: Teach-Out Plans and Agreements, Transfer of Credits, Public Disclosure and Confidentiality in the Accreditation Process, Institutional Appeals, Substantive Change, and Rights, Responsibilities, and Good Practice in Relations with Member Institutions. Eleven further edits were adopted effective immediately, including the Eligibility Requirements and the Bylaws. Comments are due Friday, December 4, 2026, by email to accjc@accjc.org, through the Commission's “Comment Now” form, or in person during the public comment session of the January 2027 Open Session. The notice ties four of the revisions to federal action: “The proposed policy revisions align with recent changes to Federal Regulations” — the Education Department's accreditation final rule at 91 FR 68626, published October 30 and also effective July 1, 2027, which we covered on November 2.
For a California community college this is the most actionable item of the week. Two things follow. First, do not finalize an institutional self-evaluation calendar for 2028 through 2032 until the implementing policy revisions land. Second, the substantive change and transfer-of-credit policies are exactly the machinery that governs how fast a college can stand up a new AI or emerging-technology program and how AI-assisted or non-traditional coursework transfers — and the notice mentions artificial intelligence nowhere. If you want AI addressed in ACCJC policy, this twenty-four-day comment window is the opening to ask for it. Read more →
CSU's faculty senate votes to point AI at paperwork, not teaching — with a Cal State Fullerton voice in it
The CSU statewide Academic Senate — 53 members representing roughly 28,000 faculty across 22 campuses — approved a resolution asking system leadership to direct AI efforts at easing faculty administrative work, such as answering email and applying grade curves, rather than promoting AI for teaching and learning. It also asks CSU to be cautious about classroom AI until there is an “informed understanding” of how it helps or hinders student learning, to support faculty in developing assessment methods suited to an AI-enabled world, and to collaborate on a systemwide study of AI's effect on student learning. It does not stop individual faculty from experimenting in their own courses, and senate resolutions carry no executive power on their own.
Jon Bruschke, a communications professor at Cal State Fullerton and a senate member, told the Sacramento Bee's Tarini Mehta that there is “so much more energy” going into encouraging AI use than into determining whether it is good for students. The backdrop: CSU's multimillion-dollar OpenAI agreement supplying ChatGPT Edu to all faculty, staff and students, and a California law passed this year — sponsored by the CSU faculty union — requiring every CSU course to have a human instructor of record. Separately, the California Federation of Teachers, representing 120,000 educators, called this week for a three-year pause on AI in K-8 classrooms.
Two disclosures. The resolution passed November 5, outside our window; the coverage is November 13, and we are reporting the coverage. And the dollar value of the CSU–OpenAI contract is still unsettled — four mutually exclusive figures are now in circulation, so we are not printing one until a CSU primary source settles it. For Fullerton College the relevance is immediate rather than abstract: CSU is where our transfer students go, the faculty member quoted works two miles away, and ASCCC resolutions track CSU senate language closely. The assessment-redesign ask is the piece a local senate can act on this spring without waiting for anyone. Read more →
Stanford screened 1,686 papers down to 52 — and found that how students use AI matters more than whether they do
Education Week's Sarah Schwartz reported on November 10 on an updated research review from the AI Hub for Education at Stanford's Accelerator for Learning. Of the 1,686 papers in the Hub's repository, only 52 tested the impact of specific interventions using randomized controlled trials or quasi-experimental designs. The review covers higher education as well as K-12, and measures academic achievement, students' perceptions of their own learning, and social-emotional outcomes. The central finding: students who used generative AI for learning support — explanations of concepts, practice problems — generally showed larger gains than students who used it to complete work, though neither use produced consistently positive outcomes across studies. Teacher use of AI improved the quantity and quality of student feedback and sometimes improved outcomes, but at least one study found it reduced student motivation and confidence. Students can think they are learning more with AI when they are not. Only 10 of the 52 studies involved U.S. K-12 participants, and the authors caution against generalizing.
A note on last week's lead. On November 9 we ran Ivy Tech professor Patrick O'Neill's audit finding that all 14 AI-learning meta-analyses he examined failed basic standards. That stands, and this review is not a rebuttal of it — the Stanford team reached past the meta-analyses to the individual trials, which is precisely the move O'Neill's audit implies you have to make. Two days later, on November 12, Harvard's Compass Collaborative launched an AI Guidance by Age matrix built from a review of more than 5,000 pieces of education, psychology and human-development research. Co-created by Jeffrey James Curley and Tessa Forshaw at the Harvard Graduate School of Education, it takes one of nine roles — “higher education educator” and “workforce educator” among them — a developmental domain and an age range, and returns recommended uses, assessment-design guidance and where AI may help or hinder a stated learning goal. Inside Higher Ed flags the obvious limit: much of that evidence base predates generative AI, some of it by decades, and Forshaw calls the matrix a starting point. Inside Higher Ed does not state whether the tool is free, and we did not confirm it.
If you teach, the support-versus-substitution distinction is the single most defensible thing to build a course-level AI policy on right now, and it is a better flex-day conversation than another tool demo. Neither tool breaks out community colleges, and the Harvard matrix's adult age ranges and workforce role are the rare AI guidance not written only for K-12. Read more →
Anthropic signs a legally binding student-privacy agreement with two teachers' unions — and it stops at K-12
On November 10, the American Federation of Teachers and the United Federation of Teachers announced that Anthropic had signed contractually enforceable terms aligned with their AI Safety and Privacy Standard for Schools, effective that day and applying to Anthropic's existing district agreements. The standard bars using student and educator data to train AI models or for advertising. As reported, the agreement limits Anthropic to using student and educator information only for authorized educational purposes, gives schools control over how long data is kept and when it is deleted, requires safeguards built for schools, and obliges Anthropic to tell schools and families what data it collects, where its products fall short and what protections are in place. Randi Weingarten, Michael Mulgrew and Anthropic's Elizabeth Kelly are quoted. The product covered is Claude for Teachers.
This also answers a question we have carried on the watch list for two weeks. Microsoft signed the same standard in September, effective November 1, and any district can add it to a new or existing Microsoft contract without renegotiating. We had been asking whether that standard reaches colleges. On the evidence published this week, it does not: both agreements are written for school districts, and neither report states a higher-education scope. In the interest of disclosure, Anthropic builds the model used in this roundup's research.
So treat this as procurement language rather than as coverage you already have. The clause set — a hard limit on what the vendor may do with institutional data, institution-controlled retention and deletion, and an affirmative disclosure duty including where the product falls short — is the most concrete public template available for a district AI contract, and a procurement officer can lift it into a request for proposals whether or not the vendor has signed anything. AFT locals represent faculty at a number of community colleges, which makes it a plausible bargaining reference too. Read more →
Major AI Companies
Microsoft announced Classroom — free on the A-SKUs, named for higher education, and not a Teams replacement
On November 11, Leif Brenne introduced Microsoft Classroom on the company's education blog as a standalone teaching-and-learning experience “designed for higher education, secondary, and primary institutions.” It consolidates Assignments, Classwork, Grades, OneNote Class Notebook, Learning Accelerators and Insights, and adds Reflect, Teach in Copilot and Learning Activities. The licensing line is the actual news: Classroom will be “available in current A1/A3/A5/A7 at no additional cost starting next year” — 2027 — and “included in A2 and above for future SKUs starting in Jan 2028.” A preview is planned for early 2027, with administrator sign-up at aka.ms/classroompreview.
The sentence that matters most to anyone running a Canvas-plus-Microsoft stack is explicit: “Microsoft Classroom does not replace Microsoft Teams for Education or Microsoft 365 LTI.” Classroom is a third option alongside using Microsoft 365 LTI inside an existing LMS, or using Teams for communication and collaboration. Microsoft published no per-student cost, because on these SKUs there isn't one, and no higher-education-specific feature breakdown.
For a community college already holding A1, A3, A5 or A7 licenses, this is a capability arriving next year with no new money attached, which is unusual enough to be worth an administrator's attention now — registering for the early-2027 preview costs nothing and buys a year of lead time on an evaluation. Two cautions. Institutions on A2 wait until January 2028, which is a real two-SKU gap inside the same system. And because higher education is named in the announcement but not detailed in it, the preview is where you find out whether “designed for higher education” means anything beyond availability. Read more →
Amazon starts paying community college costs in Virginia — last-dollar, six fields, eight colleges
The Virginia Foundation for Community College Education announced Amazon Built Together Virginia at the Governors' Luncheon in Richmond; the Virginia Community College System posted it on November 13, describing the announcement as having happened the previous day. The program covers cost of attendance, completion and certification after federal and state financial aid — a last-dollar model, so eligible students pay nothing out of pocket — and can reach fees, books, supplies, exams, certifications and required equipment. Eight colleges participate: Brightpoint, Germanna, Laurel Ridge, Northern Virginia, Piedmont Virginia, Rappahannock, Reynolds and Southside Virginia. Eligible fields are healthcare, information technology, skilled trades, education, public safety and advanced manufacturing.
Print what the release does not contain: no dollar amount, no start date and no student target. Amazon's Roger Wehner says “tens of thousands of Virginians can earn a degree or credential close to home,” which is an aspiration rather than a stated goal. One clarification on our own record: we covered Amazon's umbrella Built Together commitment on October 12 as a catch-up, and the widely quoted “more than $1 billion over the next five years” belongs to that umbrella — spanning education, job training, energy affordability, water and energy conservation and local priorities — not to this Virginia program, whose budget is unpublished. Several syndicated versions of the story are datelined November 16, which would wrongly push an in-window item out of the window; the VCCS post is the primary.
For a California community college this is precedent, not opportunity — it is Virginia-only today. But the structure is the part worth carrying into a foundation conversation: a last-dollar completion fund, routed through a system-level foundation rather than campus by campus, tied to a company's data-center community investment, and scoped to exactly the career-technical fields Fullerton already teaches. Last-dollar aid aimed at completion and certification, rather than at enrollment, is also the design most likely to move the metric colleges are actually judged on. Read more →
Adobe MAX put generative media in the Premiere timeline and previewed letting an outside agent drive Photoshop
Adobe's November 10 MAX release is the week's largest creative-tools drop. Photoshop gains Relight, Generative Restyle with an intensity slider, a Remove Tool that catches related elements like shadows and reflections, Quick Compare, and an AI Assisted Editor, with Mockups, Camera Angles, Generative Recompose, Convert to Text Layers and Timelapse in beta. Premiere makes Color Mode generally available and adds Generative Media, bringing Generate Video, Generate Sound Effects, Generate Soundscape and Generate Music into the timeline using edit context, with models from Adobe Firefly and partners including Google, Kling, Runway and Luma — plus Enhance Audio, Separate Crosstalk and Separate Sound Source. InDesign co-editing is in public beta, letting several designers work one layout in real time. Lightroom adds Glow, improved Blemish Removal, expanded tethering and Open Eyes in public beta.
Adobe also previewed Model Context Protocol integrations that let a third-party AI agent call Photoshop, Illustrator, Premiere and After Effects locally, working on native project files with layers, vectors, bins and sequences still editable; they are “coming to Creative Cloud customers in early 2027.” State the gap plainly: the post publishes no pricing, no free tier and no education licensing for any of it. For the cost the Premiere features now hide, Curious Refuge's November 12 head-to-head priced the same desert sequence through Magnific at Kling 4.0 1080p $8.40, Kling 4.0 720p $5.70, Seedance 2.5 Draft $6.00 and Seedance 2.5 Standard $24.00 — scoped to that one example, and enough to show that “generate b-roll” is a line item.
If you teach graphic design, cinema, photography or marketing, InDesign co-editing is the one that changes a workflow next week — student newspaper and yearbook production has wanted real-time co-editing for a decade. Premiere's Separate Crosstalk and Separate Sound Source double as captioning preparation, which is an accessibility win filed under video editing. And the MCP preview deserves a line in next year's academic-integrity conversation: an agent that edits native project files leaves a very different evidence trail than an exported render does. Read more →
Gemini comes to Gmail for everyone 18 and older — which leaves dual-enrollment students out
Google began rolling out Gemini in Gmail on November 10 to all Google Workspace users aged 18 and older, with users under 18 explicitly excluded. The announcement names students among the groups it is aimed at. Separately, on November 11, Google shipped the ability to publish a Sheets canvas as a standalone shareable link, reaching Rapid Release domains from November 11 and Scheduled Release domains from December 2, 2026, with availability explicitly including Google AI Pro for Education under the Education Add-ons. One sourcing note: the individual post permalinks returned errors on fetch, so both items were read from the Workspace Updates blog index, which carried the substance.
The age cutoff is the operative detail, and it is a community college problem specifically. Fullerton's dual-enrollment and early-college students sitting on under-18 accounts will not get Gemini in Gmail, which means an assignment that assumes it produces uneven access inside a single section. That is a syllabus problem rather than an IT ticket: if the tool is required, it has to be available to every student in the room, and here it provably is not. Worth checking with the district which release track it is on, too — on Scheduled Release, faculty will not see the Sheets change until December 2, three weeks after the trade press reports it as shipped. Read more →
Free & Student Opportunities
No new free AI credential launched this week — and the most-recommended one still will not quote a price
We check this every week, and the honest finding is a negative: nothing new. We swept Anthropic's academy, OpenAI Academy, Google's certificate lines, Microsoft Learn, AWS Educate and Skill Builder, NVIDIA's Deep Learning Institute, and the Coursera and edX free-audit tracks. So here is the more useful thing instead — what three of the most frequently recommended options actually cost, verified against the providers' own pages this week.
Microsoft Applied Skills: price not published. These are the credential most often described online as “free Microsoft AI certifications,” and across three of Microsoft's own pages — the Applied Skills landing page, the Applied Skills FAQ and the assessment-lab FAQ — none states a cost. What is documented is the format: you earn one “by completing interactive tasks that align to real work in a lab-based assessment,” beginner labs run 30 minutes and intermediate and expert labs 45, launching a lab counts as an attempt, and you receive “an overall passing result as well as skill level information.” The credential catalog itself renders only in a browser, so the current list of AI assessments could not be read from Microsoft's own page either. Until Microsoft prints a number, the accurate description is price not published, not free.
Google AI Professional Certificate: $49 per month. Google does print its number — “the subscription cost is $49 per month” in the U.S. and Canada, for eight courses of roughly an hour each, and “This program does not carry university credit.” Enrolling includes three months of Google AI Pro at no cost, and the page's own footnote attaches a January 1, 2027 redemption deadline to that offer. One internal inconsistency to note: a single FAQ on that page says “all seven courses” while everything else says eight. AWS Educate is the one that documents being free end to end: “completely for free,” “No credit card needed,” open to “any individual, regardless of where they are in their education, technical experience, or career journey,” learners as young as 13 with “just an email address,” and “shareable, digital badges” on completion, with the Job Board limited to learners 18 and over. What it does not claim matters too: the courses that page names are storage, compute, networking, databases and cloud operations plus a cost-estimation lab — no AI course is named there.
For a community college, the detail worth keeping is AWS Educate's 13-and-over floor, because the age-of-majority rule locks dual-enrollment students out of nearly every competition listed below, and this is one of the few things a 16-year-old on a Fullerton roster can actually start. For faculty professional development, the Microsoft assessments are still worth attempting — performance-based lab scoring is stronger resume material than a video-completion certificate — but get the price in writing from your Microsoft representative before you send a whole department after them. Read more →
AI Competitions
Deck the Pixels: $25,000, all of it cash, and no coding required
Adobe and Google Cloud are running this one through Devpost, and it is the rare AI competition built for designers rather than developers. Deadline: December 7, 2026 at 4:00 p.m. EST — 1:00 p.m. Pacific. Eligibility: above the legal age of majority in your country of residence, which is 18 in California; the main-page sidebar lists 23 excluded countries and territories including Brazil, China, Italy, Quebec, Russia and Vietnam. The page does not mention student status at all and there is no student category, so students compete in the open field. 673 registered as of November 16.
Prize, checked against the marketing: the advertised $25,000 is real and all of it is cash, with no credits or swag padding the pool — Top of the Tree (first) $8,000, Silver Bells (second) $5,000, Golden Garland (third) $3,000, The Worst Noel for best ugly sweater design $2,000 to each of three winners, Top Banana for best use of Nano Banana $2,500, and a $500 Fa-La-La Fan Favorite. That is eight separate cash awards. The task is a holiday design built in Adobe Photoshop using the Nano Banana model in three steps: an ugly holiday sweater, your own avatar or photo placed on it, and a holiday card background designed around it. You submit two final images, written answers on how you used each tool, and a screenshot of your Photoshop file with the Layers panel visible. The page states plainly: “No coding needed!”
If you teach art, graphic design, photography or marketing, this is the one to put on a board this week. It is judged on design skill, it needs no programming, the free web version of Photoshop includes a daily allowance of image generations so there is no software purchase in the way, eight cash awards against 673 entrants is unusually good odds, and three weeks is enough to run it as a class assignment rather than extra credit. Enter here →
Closing soon: four more competitions, with every prize figure checked
All four were read off the live main-page sidebar on November 16, which governs over the API and the rules tab, and every deadline is converted to Pacific.
#BuildHybridAI 2026 (Google/Firebase) — November 30, 2026 at 2:00 p.m. PST. Above the legal age of majority; Belarus, Brazil, Crimea, Cuba, Iran, Italy, North Korea, Quebec, Russia, Sudan and Syria excluded; open to individuals, teams and organizations with no student requirement. $30,000 and all of it cash: $10,000 grand prize, $7,000 and $4,000 for the runners-up, and $3,000 each for best iOS, best Android and best Web/Chrome app. 2,525 registered. You build a generative AI app on Firebase AI Logic hybrid inference, running prompts on-device where a local model exists and falling back to the cloud. Those three platform prizes are the real opening for a small team that ships well on one platform instead of chasing the grand prize.
Kiroween 2026 (AWS) — December 4, 2026 at 2:00 p.m. PST. We ran this on October 26 and again on November 2; it returns because it is now eighteen days out and because of the student bracket. Above the legal age of majority, teams up to three, sixteen countries and territories excluded including Argentina, Australia, Italy, the Philippines, Quebec and the UAE. The $100,000 headline holds up: $30,000/$20,000/$10,000 for the top three, $10,000 best startup project, $5,000 for each of four categories, and $2,000, $1,000 and $1,000 bonuses — plus four Best Student Project awards at $1,500 cash each, requiring every team member to be a student, which is $6,000 of student-only cash. The non-cash items sit on top and are named as such: student winners also get a year of Kiro Pro Max, a bonus blog prize gives 2,000 Kiro credits to each of 100 winners, and five social-blitz winners get three months of Kiro Pro+. 14,635 registered as of November 16, up from the 10,767 we printed on November 9 — date-stamp any count you quote.
QNX Global Hackathon (BlackBerry) — December 15, 2026 at 2:00 p.m. EST, 11:00 a.m. Pacific. Above the legal age of majority, no student requirement, and one caution worth stating: the main page says only “Specific countries/territories excluded” without listing them, so read the full rules before investing a semester. $50,000 in cash — $10,000 grand prize, $4,000 to the best of each of five categories, $2,000 to each runner-up, three $1,000 honorable mentions, and $3,500 each for best codelab or tutorial and best use of QNX-specific technologies. Non-cash extras are executive coffee chats, a podcast episode, a blog feature and swag. 1,651 registered. QNX is the real-time operating system running in a large share of production vehicles, and the $3,500 best codelab or tutorial prize rewards writing teaching material rather than shipping an embedded system — which is a faculty skill, not a student one.
Open Agentic Web Challenge (Cognizant with Mozilla) — January 11, 2027 at 8:00 a.m. PST. Above the legal age of majority, all countries excluding standard exceptions, no student requirement. State this prize honestly: $10,000 is cash ($5,000/$3,000/$2,000) and a further $3,000 arrives as Mozilla Otari credits stacked on top, $1,500/$750/$500 to the top three plus a $250 employee prize that carries no cash — so roughly $13,000 of value, of which $10,000 is spendable. 692 registered, the thinnest field here. Themes are machine learning and AI, web, and social good. The January 11 deadline is the only one on this list that clears finals, which makes it the one you can actually assign in a spring course that starts in January. Enter here →
Embodied AI & Robotics
Tesla will sell you a Cybercab for $29,990 — and it will not drive without a $299 monthly subscription
Tesla began inviting select customers to configure and order the two-seat, steering-wheel-free Cybercab for personal use around November 13, and posted the order agreement late that Friday; it was reported on November 14. The price is $29,990 plus a mandatory $299-per-month Autonomy Subscription, and the agreement is blunt about what that buys: “Your Cybercab will not operate without an active Autonomy Subscription.” Tesla can reprice the subscription on one month's notice. The vehicle works only inside Tesla-enabled Robotaxi geofences, which Tesla “may expand, reduce, relocate, or otherwise modify… at any time.”
The rest of the terms are the story. Tesla can remotely access, locate, restrict or control the car, place it in a “safe state,” or end the owner's access, with no remedies beyond the agreement. Autonomy Data including rear-view camera image data is retained 30 days by default. Three or more strikeouts for improper or unsafe use suspend autonomous driving for a week, with no payment proration. The car is for personal use only — putting it on Tesla's Robotaxi network, Uber, a delivery app or a rental platform requires Tesla's written authorization — and wraps, tint and wheel changes must go through Tesla or an authorized provider or the subscription can be suspended. Per our standing practice since Tesla's blog went dormant, this is sourced from press coverage of the configurator and the order agreement, not from Tesla.
There are two real teaching angles here and neither is speculative. For automotive technology programs, a production vehicle with no steering wheel, no owner-authorized aftermarket work and a remote kill switch is the clearest signal yet that independent-shop serviceability is being designed out of the product — that belongs in front of a program advisory committee this spring, not in 2028, because it bears on what a two-year automotive curriculum should still be teaching. For business, paralegal and ethics courses it is an unusually clean case study in software-gated ownership: a $29,990 asset that stops being a vehicle the month a $299 charge fails, under terms the buyer cannot negotiate. Read more →
GXO signs for up to 1,000 humanoids — on a letter of intent, not an order
On November 12, GXO Logistics signed a Letter of Intent with Apptronik to scale its Apollo humanoid under a Robotics-as-a-Service model. Joint trials begin in 2027 to validate safety, security and technical requirements; initial North American deployment is planned for late 2027; and a global rollout of up to 1,000 Apollo robots over four years is conditional on those deployments succeeding and on jointly agreed productivity, safety and operational KPIs, with each use case assessed and sequenced against them. The work covered is goods-to-person and person-to-goods picking, case and tote movement, packing and sortation. The partnership began as a 2024 proof of concept; earlier in 2026 Apollo completed a GXO pilot on autonomous picking of small consumer goods, and the two ran data collection and live autonomous evaluations across more than 50 logistics tasks at Apptronik's Robot Park training facility. GXO operates more than 1,000 facilities and over 200 million square feet with more than 150,000 team members; Apptronik has nearly 500 employees.
Print what the release does not contain: no dollar value, no named facilities, no statement that the letter of intent is binding, and no firm purchase commitment. Context from our own record: GXO expanded a separate humanoid partnership with Agility Robotics on October 26, which we covered on November 2 — so read this as a large logistics operator dual-sourcing humanoids rather than making a single bet.
For logistics and supply-chain certificate programs this is the number to teach with, and the conditionality is the honest half of it. Humanoids are not arriving in warehouses next semester; they are entering KPI-gated trials in 2027 with a four-year ramp — which lands squarely inside the working life of a student enrolling in a logistics certificate today. The service model is the career detail worth drawing out: Robotics-as-a-Service plus a “Robot Park training facility” points toward fleet-attendant, robot-technician and exception-handling roles rather than straightforward displacement. That is a credible conversation to open with an advisory committee now, while the trials are still a year out. Read more →
This roundup was assembled with AI-assisted research by the EDUAIATLAS team. Spot something we missed? Let us know.