Apple Intelligence Upgrades Outshine Siri at WWDC
ALSO: Uncle Sam eyes OpenAI, Google's Gemini Enterprise
Velkommen til Guru99 AI-rapport!
Tophistorie: Everyone’s talking about Siri’s WWDC moment, but the upgrades that’ll actually change your day slipped by quietly. Plus: Uncle Sam eyes a stake in OpenAI, and a trial collapses over AI errors. Let’s dig in.
🍎 Apple Intelligence Quietly Outshines Siri at WWDC 2026
Kort omtale:
Siri AI grabbed the headlines at WWDC 2026, but Apple’s quieter announcements may matter more. Built on its latest Foundation Models (with help from Google Gemini), a wave of Apple Intelligence upgrades hits Photos, Safari, Shortcuts, and beyond — all aimed at everyday tasks.
- Rumlig omformulering lets you drag to shift a photo’s perspective, with AI filling the gaps — joined by Forlæng and a sharper Clean Up.
- Safari auto-groups tabs by topic, mens Giv mig besked pings you when a page changes — price drops, restocks, even new menu items.
- Describe what you need and Safari builds a custom extension; Shortcuts assembles automatiseringer the same way — quietly agentic.
- Nulstilling/ændring af adgangskoder fixes compromised logins with one tap; Billedlegeplads gains photorealistic generation via Private Cloud Compute.
- Smaller wins land in Beskeder (one-tap replies), Phone (Call Context), and Home (notification summaries).
💡 Hvorfor skulle du bekymre dig?
Forget flashy chatbots — this is AI folded into apps you already use daily. When iOS 27 lands this fall, your iPhone simply gets better at small chores: rescuing photos, watching prices, untangling passwords.
🏛️ Uncle Sam Wants a Slice of OpenAI
Kort omtale:
The White House and OpenAI are reportedly in talks about the U.S. government taking an equity stake in the AI giant — with those shares potentially feeding a “Public Wealth Fund” built to hand everyday Americans a cut of the AI boom.
- Industry backers have floated a 1–5% stake, far below Senator Bernie Sanders’ proposed one-time 50% stock tax on AI labs.
- Sam Altman met with both Sanders and Trump officials last week, building on ideas from OpenAI’s April policy paper.
- Trump backed it, calling it “almost a partnership with the American public” that would “make ’em rich.”
- Former AI czar David Sacks pushed back, warning it would deepen “corporate-government fusion".
💡 Hvorfor skulle du bekymre dig?
Spreading AI wealth sounds nice. But a government that owns, profits from, and regulates OpenAI invites glaring interessekonflikter — we’ll believe it when we see it.
⚖️ Judge Cancels Trial After Both Sides File AI Errors
Kort omtale:
A Mississippi federal judge canceled an entire trial after discovering lawyers on both sides had submitted filings riddled with AI-generated errors. Senior Judge Sharion Aycock called the scenario unusual — attorneys for both litigants committing the same misconduct — and turned it into a warning shot for the profession.
- Det underliggende fee dispute between attorney Tom Withers and the city of Aberdeen derailed when both camps cited fabricated cases that don’t exist.
- All four lawyers were diskvalificeret and fined $1,000–$3,500 — two are banned for two years, and the order is headed to their state bars.
- 404 Media, which first reported the sanctions order, surfaced the judge’s blunt framing: unverified AI use reduces lawyers to a “rubber-stamp".
- Awkward timing: Anthropic just published a video on working like a lawyer with Claude — AI can organize context and draft, men judgment stays with the lawyer.
💡 Hvorfor skulle du bekymre dig?
Courts run on trust, and unverified AI is burning it on both sides of the bench. If you hire a lawyer, it’s fair to ask how they verify AI-assisted work — hallucinated citations can now sink your case.
🚨 Anthropic Sounds the Alarm — On Its Own AI
Kort omtale:
Fresh off launching its most powerful models yet, Anthropic is now waving a red flag at its own technology. The company dropped two sweeping policy proposals and a pointed essay from CEO Dario Amodei — all while racing toward a blockbuster IPO.
- Amodei’s essay, “Policy on the AI Exponential,” warns that lawmaking moves too slowly while AI races from “an amusing toy” to “a country of geniuses.”
- Advanced AI Framework demands safety transparency, independent evaluations, and government enforcement to guard against biological, cyber, and loss-of-control risks.
- Økonomisk politisk ramme maps responses to rising unemployment — from job-training grants and wage insurance to universal basic income.
- Anthropic is backing words with $ 350 millioner: $150M for a fellowship program and $200M for economic research.
💡 Hvorfor skulle du bekymre dig?
The tension: Anthropic is rare in flagging its own tech as a threat. But with a $ 965 milliarder værdiansættelse and an IPO looming, these pledges double as savvy PR.
💼 AI Agents Make Knowledge Workers More Ambitious, Not Just Faster
Kort omtale:
Perplexity teamed up with Harvard Business School researchers on a new study of how AI agents reshape knowledge work, pitting its agentic Computer platform against classic Search on 10,000 identical queries — and the results say agents change what people attempt, not just how fast they finish.
- Computer worked 26 minutter pr opgave on average versus Search’s 33 seconds — about 48x more machine effort on the same queries.
- Factor in a human doing the follow-through manually, and Perplexity estimates Search workflows take 269 minutter versus Computer’s 36 — an 87 % tidsbesparelse (and 94% on cost).
- Half of Computer requests were Create-level tasks — building docs, code, and visuals — nearly double Search’s 26%.
- Users also ventured outside their own field 59% of the time with the agent, up nine points from Search.
- Quality held up too: next-turn dissatisfaction dropped 55% compared to Search sessions.
💡 Hvorfor skulle du bekymre dig?
The headline isn’t speed — it’s ambition. When AI handles the grunt work, people don’t just finish faster; they attempt bigger, more creative projects beyond their expertise. That could quietly redraw what one person’s job can cover.
🤖 Google’s Big Bet to Tame Enterprise AI Agents
Kort omtale:
Google Cloud Next ’26 just dropped its headliner: Gemini Enterprise, an end-to-end platform built to help businesses build, manage, and deploy AI agents at scale. Think one secure hub where your apps, data, and autonomous helpers finally play nice together.
- At its core sits the Gemini Enterprise Agent-platform — an evolution of Vertex AI bundling Google’s full model suite with new Hukommelsesbank og Agentidentitet tools that give IT teams real control over their agent fleet.
- Agentdesigner lets anyone build agents with ingen kode, using plain language or a visual interface.
- Projekter, Canvas, and an Inbox keep teams and agents collaborating, with real-time alerts on agent behavior.
- En åben Agent Gallery offers partner-built agents, while BYO-MCP forbinder Gemini to your other business tools.
💡 Hvorfor skulle du bekymre dig?
Drowning in disconnected AI tools? Google’s pitch is order: agents that remember context, respect security rules, and live where you already work. The catch — you’re betting deeper on Google.
Hej! Jeg er Krishna Rungta
Grundlægger af Guru99.com, chefredaktør og teknologiekspert
Blev denne e-mail videresendt til dig? Tilmeld dig gratis her.
