Graduating Seniors Boo Eric Schmidt Every Time He Mentions AI in Commencement Speech

01Eric Schmidt Tried to Sell AI to Graduating Seniors. They Booed Every Time He Said It

Eric Schmidt stood at the University of Arizona commencement on Friday and began talking about artificial intelligence. The boos started. He kept going. The boos came back. According to The Verge, the former Google CEO was "repeatedly drowned out" as his speech veered into AI cheerleading, with the loudest jeers landing on the sections meant to inspire.

The students booing him were minutes away from receiving diplomas and walking into the job market that AI is currently rearranging. Entry-level software roles are the first cohort being absorbed by coding assistants. Marketing, paralegal, and customer service pipelines are thinning. Schmidt, who left Google in 2017 and has spent the years since promoting AI as civilizational progress, picked a venue full of people who will discover within months whether his framing survives contact with a hiring manager.

He is not the only speaker to have miscalculated. TechCrunch, which covered the Arizona incident the same day, ran a separate piece on Saturday with a blunt suggestion for the rest of commencement season: if you are giving a graduation speech in 2026, maybe don't mention AI. The publication's reasoning was that it has become "tough to get graduating students excited about a future shaped by artificial intelligence."

The reaction at Arizona was not generalized anti-tech sentiment. Schmidt was not booed when he opened his remarks. He was booed on a specific subject, at a specific moment, by an audience whose proximity to the consequences of that subject is measured in weeks. The optimism in his script and the position of his listeners did not line up.

Schmidt's recent public commentary has leaned into the same register. He has warned about Chinese AI capabilities, called for massive infrastructure spending, and described the technology as the defining force of the coming decade. None of that framing addresses the question the people in caps and gowns were asking, which is whether the labor market will have a place for them once the model rollouts finish. He offered them a future. They asked, audibly, who it was for.

The class of 2026 will sit through dozens more commencement speeches over the next three weeks. TechCrunch's advice will likely be ignored.

02Apple's pitch for the new Siri: an AI assistant that forgets you on purpose

The chatbot-style Siri arriving in iOS 27 will let users auto-delete their chat histories, according to a Bloomberg report from Mark Gurman cited by The Verge and TechCrunch. Apple is treating data disposal as a headline feature of the revamp, not a buried setting.

The framing inverts where the rest of the assistant market is heading. OpenAI this month wired ChatGPT into users' bank accounts through a new connector layer, expanding the surface of personal data the model can read. Google updated its search spam policy the same week to treat coordinated AI-driven manipulation as a violation, an acknowledgment that assistants are now active agents inside its index. Anthropic, Perplexity, and Microsoft Copilot all retain conversation histories by default to power memory and personalization features their roadmaps depend on.

Apple is going the other way on the one product where it most needs to catch up. Siri has trailed ChatGPT, Gemini, and Claude on reasoning, coding, and multi-turn dialogue for the better part of three years, and the iOS 26 Apple Intelligence rollout did not close the gap. Rather than match competitors on capability, the company is according to Gurman's reporting building its pitch around what it will not keep.

The trade-off is real. Assistant memory is what makes ChatGPT remember a user's writing style, what lets Gemini recall an ongoing project, what makes Copilot useful across a workweek. Auto-deletion strips that out. A Siri that forgets every conversation is a Siri that cannot personalize across sessions, which is precisely the axis on which rivals are competing hardest.

Apple has run this play before. App Tracking Transparency in 2021 cost Meta an estimated $10 billion in ad revenue in its first year and reset expectations for what mobile operating systems owe their users on data. The company is betting the same instinct still moves iPhone buyers when the product in question is a chatbot rather than an ad network.

What is not yet public, according to The Verge and TechCrunch, is the default state of the setting, the retention window before deletion, or whether server-side copies persist for safety review. Apple has not detailed any of those mechanics. The reveal is expected at WWDC next month.

03Malta bought ChatGPT Plus for every citizen. Enterprises can't justify their seats. Tahoe ratepayers got the bill anyway.

OpenAI and the government of Malta announced a partnership to provide ChatGPT Plus to every Maltese citizen. The company described it as the first national rollout of premium consumer AI access. No other country has signed such a contract publicly with OpenAI. Neither side disclosed the per-seat price, the contract term, or how usage would be metered. The arrangement reclassifies consumer AI: a treasury, not a household, writes the check. What was a direct-to-consumer subscription becomes a government procurement line.

Inside companies, the same bill refuses to settle. A Hacker News post titled "AI subscriptions are a ticking time bomb for enterprise" drew 364 points. It argued per-seat AI licensing creates renewal exposure and compliance audit gaps unfamiliar to traditional SaaS contracts. The harder problem is internal accounting: IT cannot reliably tell which assigned seats are being used, by whom, or for what kind of work. Companies pay the invoice and cannot price the value, defend the spend, or trim the waste at renewal. The result, the post argued, is a recurring cost line that neither finance nor security can fully audit.

Further from any subscription line item, the cost arrives as electricity. TechCrunch reported that residents around Lake Tahoe, long Silicon Valley's vacationland, are about to face higher rates as AI-driven data center demand pressures the regional grid. The area is now looking for a new energy provider in response to the strain. Ratepayers absorbing the increase did not sign up for anything labeled AI; their bill is going up because someone else's model training is.

Three settlements in one week, three different parties holding the bill. The treasury wrote a check on behalf of its citizens. In the enterprise, a CFO signed one with no internal accounting for what it bought. Households near data centers absorbed one they never agreed to. None of the three is the retail user OpenAI markets ChatGPT Plus to.