1. Meta Begins Cutting 8,000 Jobs Today — 10% of Workforce Gone
Meta starts laying off approximately 8,000 employees today, May 20, as the $145 billion AI spending plan reshapes the company from the inside. Another 6,000 open roles are cancelled. And more cuts are coming — sources say August and later this year.
This is the third major round since Zuckerberg’s “year of efficiency” in 2023, but the tone has shifted dramatically. There’s no apology this time. The memo reads like a spreadsheet decision: “all part of our continued effort to run the company more efficiently and to allow us to offset the other investments we’re making.”
Finance chief Susan Li told analysts executives “don’t really know what the optimal size of the company will be in the future.” Which is a hell of a thing to say while firing 8,000 people.
Why it matters: The 2026 tech layoff pace is approaching 2023 levels — 110,000 cuts so far this year. But the justification has changed. In 2023 it was “we overhired during COVID.” In 2026 it’s “AI lets us do more with less.” That’s a structural shift, not a correction. These jobs aren’t coming back.
2. Musk Loses OpenAI Lawsuit — Jury Says He Waited Too Long to Sue
A California jury unanimously dismissed Elon Musk’s lawsuit against OpenAI and Sam Altman, ruling he exceeded the statute of limitations. The verdict came down May 18 in Oakland.
Musk called it a “calendar technicality” on X and vowed to appeal. The jury didn’t reach the substance — whether OpenAI abandoned its nonprofit mission or violated antitrust laws — because the statute of limitations clock ran out before Musk filed.
The trial had already delivered some of the year’s best courtroom moments. Musk admitted under oath that xAI distills OpenAI’s models while claiming to be AI safety’s guardian. The courtroom gasped. That admission — that the company suing OpenAI for betraying its mission is literally built on OpenAI’s technology — was worth the trial all by itself.
Why it matters: The appeal will likely fail on the same procedural grounds. But the distillation admission is now on the public record and can’t be un-said. It’s the kind of fact that lives in every future antitrust discussion about AI training data and model weights. Also, Musk v. Altman is now 1-0 to Altman, and it got personal enough that neither side walks away clean.
3. Pope Leo XIV to Launch First AI Encyclical With Anthropic Co-Founder
Pope Leo XIV will publish Magnifica Humanitas (“Magnificent Humanity”) on May 25 — his first encyclical, focused entirely on AI and human dignity — with Anthropic co-founder Christopher Olah present at the Vatican launch.
The symbolism is layered. The encyclical is signed on May 15, exactly 135 years after Pope Leo XIII’s Rerum Novarum — which addressed workers’ rights and the limits of capitalism during the Industrial Revolution. The current pope is deliberately drawing the parallel: AI is this century’s industrial revolution.
Anthropic’s presence is a geopolitical signal. The Vatican chose the safety-first AI company at a moment when Anthropic is actively suing the Trump administration for retaliation over refusing military AI access. Olah isn’t just there to talk technology — he’s there because the Vatican is picking a side.
The launch features two top cardinals, theologians, Anthropic’s co-founder, and the pope himself. Unusually high-profile for what’s normally a press room event.
Why it matters: A former math major as pope, writing the church’s first major AI doctrine, at a moment when the leading AI safety company is in a legal war with the US government. The timing is not coincidental. For New Zealand, where the government still has no binding AI framework, an international moral authority laying out AI ethics principles carries weight — even in a secular society.
4. YouTube Opens AI Deepfake Detection to All Adult Users
YouTube’s “likeness detection” tool — which scans uploaded videos for AI-generated content using a person’s face — is now available to all creators over 18. Previously limited to high-profile figures, the tool lets anyone monitor content using their likeness and request removal of deepfakes.
The technology uses a selfie-like scan to create a biometric signature, then continuously checks new uploads against it. YouTube says it catches face-swapped content even when the original context is completely different.
Why it matters: Deepfake porn and impersonation have been a growing crisis on every platform. Making this tool universal is the right move — but it also creates a centralised biometric database controlled by Google. There’s always a trade-off. For NZ creators, this is a genuine protection tool. For NZ privacy advocates, it’s more questions about data sovereignty.
5. Ken Griffin Called AI “Garbage” in January. Now He’s “Depressed.”
Citadel CEO Ken Griffin, one of the most prominent AI skeptics in finance, has completely reversed his position after seeing what the technology can actually do inside his own firm.
At Davos in January, Griffin told a panel AI was “garbage” the moment you dug beneath the surface. In May, speaking at Stanford Business School, he described going home “actually fairly depressed” after watching AI agents at Citadel complete work in hours that previously took teams with master’s degrees and PhDs weeks or months.
The conversion happened fast. Griffin now says “the success in your career will be defined as to whether or not you will be a lifelong learner” — and that AI’s impact on high-end research is “quite eye-opening.”
Why it matters: When the CEO of one of the world’s most profitable hedge funds — a man famous for being right about markets — does a 180 from “garbage” to “depressed” in four months, it’s worth paying attention. He’s not talking about chatbots. He’s talking about knowledge workers with advanced degrees being replaced by agents. The “safe” jobs are the ones getting automated now.
6. OpenAI Merges ChatGPT and Codex Under Greg Brockman — Side Quests Are Over
OpenAI has permanently consolidated ChatGPT, Codex, and its developer API into a single product organisation led by co-founder Greg Brockman. The restructuring kills the company’s “side quests” — Sora (video generation), OpenAI for Science, and the adult mode for ChatGPT were all shelved — as the company prepares for its Q4 2026 IPO at an estimated $852 billion valuation.
The merger means one unified “agentic platform” rather than separate consumer and developer products. OpenAI told investors (implicitly) that maintaining multiple product lines while resource-constrained was a luxury it could no longer afford. Brockman himself said on a podcast that OpenAI’s computing power is “not enough for even a personal assistant and the Codex line.”
The competitive context is unforgiving. Cursor hit $2 billion ARR and is valued at $50 billion. Google’s Gemini grew from 5.7% to 21.5% of AI web traffic in 12 months while ChatGPT’s share dropped from 86.7% to 64.5%. OpenAI is responding with an org chart, not a product launch.
Why it matters: The IPO is the driving force. A single “super app” narrative is easier to pitch to institutional investors than a portfolio of experiments. But this also means OpenAI is narrowing its bets at exactly the moment when competitors are diversifying. If the super app thesis is wrong, there’s no plan B.
7. Amazon Alexa+ Can Now Generate Podcast Episodes
Amazon’s latest Alexa+ update lets the assistant generate full podcast episodes on demand. The feature creates podcast-style audio content on any topic, with AI voices and production quality that Amazon says is indistinguishable from human-created podcasts.
The feature is the latest example of AI moving from text generation into audio content production. It joins NotebookLM’s podcast-style audio summaries and Google’s Illuminate as signs that synthetic audio is becoming a commodity capability.
Why it matters: The podcast industry — which spent years building audiences on human connection and authentic voices — now faces the same disruption as every other content medium. If AI can generate an indistinguishable podcast episode on any topic in seconds, the economics of human podcasting change fundamentally. NZ’s podcast scene, still relatively small, should be paying attention.
8. OpenAI Partners With Dell to Bring Codex Into Enterprise
OpenAI has signed a partnership with Dell to integrate Codex into enterprise AI systems, connecting the coding agent with Dell’s AI Data Platform and Dell AI Factory for hybrid and on-premises deployments.
The deal is significant because it signals enterprise demand for Codex beyond the developer tools market. Companies want AI coding assistants deployed inside their own infrastructure, not just in the cloud. Dell’s reach into corporate IT departments — still the dominant channel for enterprise technology procurement — gives OpenAI distribution it couldn’t build alone.
Why it matters: Enterprise AI deployment is moving from “let’s try ChatGPT” to “let’s embed AI agents in our internal systems.” Dell is the infrastructure play. This partnership, combined with the Brockman restructuring, suggests OpenAI is serious about the enterprise — but so is everyone else.
9. Detroit’s Big Three Have Cut 20,000 White-Collar Jobs — and AI Is Accelerating It
General Motors, Ford, and Stellantis have eliminated 19% of their white-collar workforce since 2022 — approximately 20,000 jobs — with AI-driven automation now accelerating the pace.
GM’s latest move tells the story: the company laid off 500 IT workers this week while simultaneously posting 250 new AI-focused roles. Same company, same quarter. The “swap” narrative — lose your job to AI, get reskilled into an AI job — is being tested in real time, and the math doesn’t work. 500 gone, 250 hired. That’s not a pivot, it’s a contraction.
Why it matters: This is the pattern playing out across every industry. The “reskilling” story assumes the new jobs replace the old ones at the same rate. Every data point we’ve seen this year says otherwise. The ratio is closer to 2:1 or 3:1 in favour of elimination. Career Compass will have more on this tomorrow.
10. Dust Raises $40M to Make AI “Multiplayer” Inside the Enterprise
Dust, an enterprise AI platform, raised $40M in Series B funding from Abstract and Sequoia to build what it calls “multiplayer AI” — tools that let entire teams collaborate through AI agents rather than individual one-on-one chatbot sessions.
The thesis is that enterprise AI adoption has stalled because current tools are designed for individual use. Dust wants AI that sits inside a company’s workflow, accessible to everyone on a team, with shared context and persistent memory.
Why it matters: “Multiplayer AI” is the next logical step after individual AI assistants. If Dust is right, the enterprise AI market shifts from “each person uses their own chatbot” to “the whole team uses one shared AI.” That’s a fundamentally different product category — and one that incumbents like OpenAI and Microsoft haven’t fully addressed.
🔍 THE BOTTOM LINE
May 20 is a day of two stories running in parallel.
In one story, the AI industry is consolidating, IPO-ing, building encyclicals with the Vatican, and making podcast episodes from nothing. OpenAI merges its products and eyes $852 billion. The Pope launches Magnifica Humanitas. YouTube hands every adult a deepfake detector. Ken Griffin, the AI sceptic, goes home depressed because his firm’s agents are doing PhD-level work in hours.
In the other story, 8,000 people at Meta are clearing their desks today. GM fires 500 IT workers and rehires 250. Detroit’s Big Three are down 20,000 white-collar jobs and it’s accelerating. The ratio is never 1:1. The reskilling story doesn’t hold up to arithmetic.
These are not separate stories. They are cause and effect. The AI industry’s most impressive achievements are also somebody’s redundancy notice. The question is not whether the technology works — Griffin’s conversion proves it does. The question is what happens to the people whose jobs were the “optimisation” line on a spreadsheet.
The Pope is asking this question. The markets are not.
This is a daily news digest from Singularity.Kiwi. Full analysis articles on major stories published separately.