For University Faculty & Staff · ~10 min read · June 2–9, 2026 · ET Newsletter Home
Intelligence Briefing · No. 5

AI in
Education

Both major AI labs filed to go public within a week. Berkeley's CS classes hit a failure-rate wall. And the university officer in charge of integrity got caught using AI to write about integrity. Higher ed's vendor and trust posture is changing in real time.

2
Major AI labs filed for IPO this week
$965B
Anthropic's pre-IPO valuation
30
Berkeley CS 10 students caught cheating
63%
K-12 teachers saving ≤2 hrs/week with AI
Issue briefing video

Watch the Issue 5 briefing.

A short video companion to this AI in Education issue, placed here before the section navigation so readers can watch before diving into the full briefing.

Five numbers,
one week.

If you read nothing else, read these. Each card answers a question a colleague will ask on Monday.

⚖️ Policy · Higher Ed
1st

Western Sydney's Pro Vice-Chancellor for Quality and Integrity was caught using Microsoft Copilot to write an op-ed warning students against cutting corners. The university defended the use as “sophisticated and appropriate.” The Sydney Morning Herald retracted the piece.

🛠️ Tools · 🏛️ Policy
2

Anthropic filed for IPO on June 1 at a $965B private valuation. OpenAI filed on June 8. Higher-ed AI contracts are now implicit bets on public-company earnings calls.

🎓 Higher Ed
30

Berkeley teaching prof Dan Garcia reports ~30 students in CS 10 alone were caught cheating on take-home exams in spring 2026, citing “a vast increase in academic dishonesty” tied to ChatGPT, Claude, and Gemini.

⚖️ Policy
3rd

The House Subcommittee on Higher Education held its third hearing in a deliberate AI series on June 3 — “Building an AI-Ready America.” Federal legislation is now a probability.

📊 Research
63%

NPR/Ipsos poll of K-12 teachers: most who use AI say it saves them two hours per week or less. Majority worry it's making it harder for students to think for themselves.

When the rules apply
to everyone except the rule-makers.

The feature story is a small one in the grand scheme. But it captures exactly the standard your campus is about to have to articulate.

A Pro Vice-Chancellor for Integrity used Copilot to write an op-ed warning students against AI. Pangram caught it.

Prof. Cath Ellis — Western Sydney University's Pro Vice-Chancellor for Quality and Integrity — published an opinion piece in The Sydney Morning Herald telling students “don't cut corners” and “don't outsource your thinking.” The Guardian ran the piece through Pangram, which flagged it as AI-generated. WSU confirmed Ellis had uploaded 40,000 words of her own materials into Microsoft Copilot to produce the draft, and the university media team had then run a second AI pass before submission. None of it was disclosed to the SMH.

The SMH retracted the piece. Western Sydney's spokesperson called the use “sophisticated and appropriate” and consistent with the university's “institutional position of human-centred AI.” Ellis maintains she wrote the piece with AI, not by it. That distinction — and who gets to draw it — is exactly the standard your campus is about to have to articulate, for students and for itself.

Guardian, June 3 →   SMH retraction →   ABC News →
What this means for your campus

Your AI policy needs an explicit faculty- and staff-facing disclosure rule for external publication. Otherwise the standard you ask students to follow doesn't apply to the people enforcing it — and they will notice.

OpenAI files for IPO one week after Anthropic. Both AI labs head to public markets.

Anthropic filed a confidential S-1 on June 1 at a $965B private valuation. OpenAI followed on June 8, recently valued at $852B. Both companies are unprofitable. Reuters and TechCrunch note Anthropic's filing will set a valuation comp that constrains OpenAI's pricing. Before signing a three-year campus AI contract this summer, ask procurement whether your termination clause survives an acquisition, a pricing reset, or a change in alignment policy.

Reuters →   TechCrunch →

UC Berkeley CS classes hit a failure-rate wall — faculty cite a “vast increase in academic dishonesty”

Berkeley teaching professor Dan Garcia (CS 10, CS 61A) told the Daily Californian that “a primary driver” of unusually high failure rates last semester was AI-assisted cheating. ~30 CS 10 students were caught on take-home exams in spring 2026. Reads as the in-classroom counterpart to the Chirikov grade-inflation paper: the students who get caught fail, the students who don't inflate.

Times of India recap →

House Subcommittee holds third hearing in deliberate AI-in-higher-ed series

Rep. Burgess Owens chaired “Building an AI-Ready America: Higher Education in the Age of AI.” Michael Horn (Harvard GSE) used his testimony to call for accreditation reform and outcomes-focused metrics. Legis1 reports federal AI-readiness legislation is now a probability. Your government-relations office should be watching the subcommittee markup calendar through the summer.

Horn's testimony →   Legis1 analysis →

EDUCAUSE Horizon Report 2026: trust is now the binding constraint, not policy

The 2026 EDUCAUSE Horizon Report names a growing “trust and credibility” gap between faculty and students as AI's central risk to higher ed. The report cites a Google experiment with AI-generated, per-student textbooks as an emerging signal, and argues institutional governance will need systems that don't rely solely on human oversight. Use the Horizon framing for your next senate or board presentation — “trust gap” is the language being adopted across the sector. Coverage in GovTech, June 2.

Talk about it

If our institution's most senior integrity officer used AI to write an op-ed without disclosure, would our faculty handbook treat it as misconduct or as “sophisticated use”? Should it depend on the venue?

Talk about it

Both major AI labs will be public companies within months. What is the appropriate term length for an institutional contract with a pre-IPO AI vendor?

What scholars learned.
What faculty built.

Two columns: studies that should shape how you think; two real institutions that should shape what you do.

From the Research

Annual report · Delphi method

EDUCAUSE Horizon 2026: AI is straining the student-faculty trust relationship

Annual synthesis from a global panel of higher-ed experts. The 2026 edition identifies a growing “trust and credibility” challenge as AI becomes embedded in coursework, alongside a call for governance systems that don't rely solely on human oversight.

What you could do: use “trust gap” as the framing for your next senate or board presentation. The vocabulary is now sector-standard, so peers will recognize it.

GovTech summary →
National poll · K-12

NPR/Ipsos: most teachers using AI save ≤2 hours/week — and worry students can't think for themselves

A nationally representative poll of K-12 teachers conducted by Ipsos for NPR finds 63% of AI-using teachers report saving two hours per week or less, while a majority worry the technology is making it harder for students to learn to think for themselves.

What you could do: use this against marketing claims of dramatic teacher-productivity ROI when you're sitting in a vendor pitch. The classroom data does not support the slideware.

NPR, June 5 →

How Others Are Doing It

Workforce pipeline · MIT + GSU

MIT RAISE + Georgia State launch PATH for community-college AI training

Cynthia Breazeal's MIT RAISE group, with Georgia State and Google.org backing, announced PATH (Pathways for AI Training and Hiring) — a multiyear initiative explicitly focused on transforming community colleges into the AI-workforce engine. Connects research universities, community colleges, and industry into a shared pipeline.

What you could borrow: if your university has a community-college transfer pipeline, MIT just published a structure for sharing AI curriculum across that boundary instead of recreating it course by course.

MIT News, June 4 →
Program redesign · Spain

IESE rebuilds the MBA around AI in every first-year course

Deputy dean Marc Badia and the IESE faculty, with Wharton's Ethan Mollick as outside thinker, embedded AI into every Y1 MBA course rather than bolting it onto a couple of electives. The pivot frames AI's challenge as identity reskilling, not skill reskilling. Badia: “We made a deliberate choice not to bolt AI onto a couple of electives and call it a transformation.”

What you could borrow: even one program-level rebuild — not a faculty-by-faculty optional add — gives your dean a real story to tell admitted students this fall.

Poets & Quants, June 2 →

Everything else worth
knowing this week.

Eight stories that didn't earn full commentary — and four dates that should be on your calendar.