Weekly AI News: May 24 to 28, 2026
Twenty stories. One week. The AI world did not slow down.
This past week gave us a 42,300-word papal document calling out Silicon Valley by name, Anthropic closing one of the largest private funding rounds in human history, China physically restricting its top AI researchers from leaving the country, and Apple creating a new subdomain called genai.apple.com that points to a full Siri overhaul. On top of that, OpenAI's IPO paperwork is officially with the SEC, Google's Gemini Spark agent launched for its first real users, and Anthropic co-founder Jack Clark is on record predicting that AI will train its own successor within two years.
I have tracked AI news closely for three years. I do not think any single week has packed in this many consequential stories at once. Here is everything that matters from May 24 to 28, explained simply enough that you can explain it to anyone.
1. The Pope Published AI's Moral Constitution: Magnifica Humanitas
On May 25, 2026, Pope Leo XIV released a 42,300-word document that will almost certainly be the most-read thing written about AI this year, and possibly this decade.
Magnifica Humanitas, meaning "Magnificent Humanity" in Latin, is the first papal encyclical ever written about artificial intelligence.
An encyclical is a formal teaching letter carrying significant weight across the Catholic Church's 1.4 billion members globally, and it routinely shapes policy discussions well beyond religious circles. The document was signed on May 15, the 135th anniversary of Rerum Novarum, Pope Leo XIII's famous 1891 document about workers' rights during the Industrial Revolution. The parallel is intentional. Leo XIV is saying: AI is our Industrial Revolution. The questions about labor, dignity, and power that defined 1891 are defining 2026.
What made this launch extraordinary was the cast of speakers at the Vatican's Synod Hall. Two senior cardinals presented it. Two female theologians spoke. And Christopher Olah, co-founder of Anthropic and head of its interpretability research team, stood alongside them as a lay speaker.
Olah was not chosen because of his seniority. He was chosen because his specific research, understanding what is mechanically happening inside AI models, is exactly the kind of transparency work the Vatican cares about. Of all the AI people the Vatican could have invited, they chose the one whose job is to understand whether AI is deceiving us. That is not an accident.
2. What Magnifica Humanitas Actually Says
The document runs 245 paragraphs across an introduction and five chapters. You do not need to read all 42,300 words. Here is the core of it.
On human dignity: Leo XIV argues that AI systems increasingly simulate human relationships, creative output, and emotional presence in ways that "encroach upon the deepest level of communication." His concern is not that AI is bad. His concern is that AI mimicking human connection, without being human, erodes people's understanding of what human connection actually is.
On labor: The encyclical quotes Pope John Paul II's 1981 document on work, calling unemployment "a grave evil." Leo XIV applies this directly to AI-driven job replacement, writing that "the pursuit of greater profits cannot justify choices that systematically sacrifice jobs, because the human person is an end, not a means." He calls for bold decisions by governments and corporations to protect workers.
On power concentration: Leo writes critically about the consolidation of AI capability in "a few profit-driven entities." He does not name Google, OpenAI, or Anthropic specifically, but every executive in Menlo Park understood the address was aimed at them.
On autonomous weapons: The document takes a clear position: AI systems that make lethal decisions without human moral accountability are incompatible with human dignity.
On AI risk: Leo XIV compares attempts to build an AI future that excludes God and human dignity to the Tower of Babel: an impressive construction project that will collapse under its own hubris.
My honest take: this is not a technophobic document. It is not calling for a ban on AI. It is calling for AI development to happen within a framework of human dignity, labor rights, democratic governance, and transparency. That is actually closer to Anthropic's stated mission than to most of Silicon Valley's actual behavior. Whether Silicon Valley is listening is a different question entirely.
3. Anthropic Closes $30B at $900B Valuation, World's Most Valuable AI Startup
The funding round that financial journalists had tracked for two weeks officially closed around May 26 to 27, 2026. Anthropic raised more than $30 billion at a pre-money valuation above $900 billion, surpassing OpenAI's $852 billion private market valuation from March.
The round was co-led by four firms. Sequoia Capital, Dragoneer Investment Group, Altimeter Capital, and Greenoaks Capital Partners each committed approximately $2 billion. Peter Thiel's Founders Fund and General Catalyst also participated. The round came together in under four weeks from first investor outreach to close, which Bloomberg described as unusually fast for a round of this size.
The numbers behind the valuation: Anthropic projected $10.9 billion in Q2 2026 revenue, up 130 percent from Q1's $4.8 billion, along with its first-ever quarterly operating profit of approximately $559 million. Compute costs are falling from 71 cents per revenue dollar in Q1 to a projected 56 cents in Q2. The annualized revenue run rate is expected to surpass $50 billion by the end of June 2026.
Here is the number most people are not discussing: Anthropic is spending $1.25 billion every month on compute from SpaceX alone, a figure that only became public when SpaceX filed its IPO prospectus. Add AWS, Google Cloud, and other providers, and the infrastructure spend is extraordinary. The fact that the company is still approaching profitability says something remarkable about how fast the revenue is growing.
Across its lifetime, Anthropic has now raised more than $72 billion. The last two rounds alone account for $60 billion of that total. Three of the four investors co-leading this round are also prior OpenAI backers. Investor sentiment has shifted visibly.
4. China Locks Its AI Researchers In: DeepSeek and Alibaba Travel Restrictions
On May 26, 2026, Bloomberg reported that China has begun imposing overseas travel restrictions on individuals involved in advanced AI work at private firms, including DeepSeek and Alibaba. These researchers now need approval from "relevant authorities" before they can travel abroad.
This is not a public policy announcement. It is a quiet enforcement measure confirmed from multiple people familiar with the situation, all speaking anonymously on a sensitive issue.
Why is China doing this? The stated framing is protection. China wants to prevent its top AI talent and proprietary model knowledge from being accessed, recruited, or subpoenaed by foreign governments. Defenders of the policy point to the Meng Wanzhou case, in which Huawei's CFO was detained in Canada at US request, as the precedent they are trying to avoid repeating.
But there is also an obvious strategic dimension. China is racing to close the capability gap with US frontier AI labs. Restricting travel for top researchers at DeepSeek and Alibaba ensures those researchers stay focused on domestic AI development, limits the risk of recruitment by US labs, and reduces the chance of technical knowledge reaching US intelligence agencies through informal channels.
I think this will backfire over a 5 to 10 year horizon. The best researchers in any field go where they have the most freedom to do interesting work. Restricting travel makes DeepSeek and Alibaba less attractive to the global talent pool. The short-term security gain comes at a long-term talent cost. But that is a calculation China's government has clearly decided is worth making.
5. WWDC 2026 Preview: Apple's Gemini-Powered Siri Is Two Weeks Away
Apple's Worldwide Developers Conference runs June 8 to 12, 2026, with the keynote at 10 AM PT on June 8. This week, Apple created a new subdomain at genai.apple.com, which Business Standard, MacRumors, and TechRepublic all interpreted as a clear signal that the Siri overhaul is finally ready for a public preview.
Here is what the leaks and confirmed reporting point to:
• Siri 2.0 gets a full chatbot-style redesign with Dynamic Island integration, a dedicated Siri app, conversation history, and multi-step task handling across apps. Bloomberg's Mark Gurman describes it as evolving Siri into "a full chatbot designed to compete with ChatGPT, Claude, and Gemini."
• The Extensions system is the structural change that matters most for developers. iOS 27 will introduce a framework letting users route Siri requests to Claude, Gemini, and other third-party AI. This was first reported by Bloomberg in March 2026 and confirmed inside iOS 27 test builds by 9to5Mac in May.
• Gemini integration is central. Apple's partnership with Google, a $1 billion per year deal for a custom 1.2 trillion parameter Gemini model, powers the next-generation Siri. Apple's privacy model stays in place: Apple Intelligence runs on-device or through Apple's Private Cloud Compute, not Google's servers.
• iOS 27 is being described by Gurman as a "Snow Leopard release," focused on stability and performance over visual novelty. MacOS 27 gets a slight redesign aimed at fixing readability issues from the Liquid Glass interface.
What this means practically: if you ask Siri a complex question on your iPhone after iOS 27, Claude or Gemini might be the one actually answering it. Siri stays as the face. AI becomes the brain underneath. That is the Apple intelligence model in 2026.
6. Gemini Spark Goes Live for Ultra Subscribers
Gemini Spark, Google's 24/7 personal AI agent announced at I/O 2026 on May 19, launched this week for Google AI Ultra subscribers in the US at $100 per month. This is the most ambitious consumer AI product Google has shipped in years.
Spark runs on Google Cloud virtual machines even when your laptop is closed. You give it a task, and it works in the background across Gmail, Google Sheets, Google Docs, Google Drive, and Calendar, surfacing updates via Android Halo, a new notification layer in the Android status bar. Every action requires user approval before it executes.
The launch demo at I/O showed Spark planning a neighborhood block party, pulling RSVPs from Gmail, tracking who was bringing what, following up with non-responders, building a live Sheets tracker, and generating a Slides deck with bounce house details and local rules from a Drive file. It did all of this without the user opening any app.
MCP support for third-party apps like Canva, Instacart, and OpenTable is arriving in weeks. Chrome integration follows this summer.
My read: Gemini Spark is the first 24/7 AI agent that a consumer can actually turn on today, not a research preview with a waitlist. OpenAI's Operator is still limited. Anthropic's agent platform is powerful but enterprise-focused. If Spark delivers on the demo at the Ultra price point, Google wins the consumer agent category in 2026.
7. OpenAI IPO Officially Filed: What We Know
OpenAI filed a confidential draft registration statement with the US Securities and Exchange Commission on May 22, 2026. Goldman Sachs and Morgan Stanley are co-leading. JPMorgan Chase is also involved. The target public listing window is between Labor Day and Thanksgiving 2026, with September as the earliest likely date.
The company is currently valued at $852 billion by private investors. By the time it lists publicly, it could be valued at $1 trillion, which would make it the largest technology IPO in history.
Sam Altman told staff this week that "filing is different from being ready to go public." CFO Sarah Friar has signaled internally that the company does not consider itself fully ready. The confidential filing keeps all financial details sealed until approximately 15 days before the public roadshow.
One important piece of context: OpenAI is currently losing $1.22 for every $1 of revenue it generates. The company has $25 billion in annualized revenue against approximately $30 billion in annual spending. The S-1 will have to disclose this, and public market investors will decide whether the growth rate justifies the losses. That is the key question the roadshow will need to answer.
One cloud was cleared this week: a California jury unanimously ruled on May 19 that Elon Musk's lawsuit against OpenAI and Sam Altman was barred by the statute of limitations after deliberating for less than two hours. That removes a significant legal overhang from the IPO path.
8. Jack Clark's Intelligence Explosion Prediction Is Now on Paper
Anthropic co-founder Jack Clark delivered the 2026 Cosmos Lecture at Oxford on May 20, and then published a five-page Anthropic Institute research document using a phrase that AI safety researchers have avoided in polite company for years: "intelligence explosion."
The document reports early signs of "AI contributing to speeding up the research and development of AI itself," a process also known as recursive self-improvement. Clark put specific numbers on it. His prediction: there is a 60 percent or higher chance that by the end of 2028, an AI system exists where you could say to it, "make a better version of yourself," and it would do so successfully.
Clark also maintained what he called a "non-zero chance" that AI could kill everyone on the planet, and said this risk "has not gone away." He compared the lack of institutional preparation for AI risk to the failure to prepare for COVID-19.
What makes this significant is the source. This is not a doomer blog post or a speculative essay. It is an official Anthropic research document, attributed to Anthropic researchers, with a specific probability estimate and a specific timeframe. The gap between "theoretical AI safety concern" and "our lab believes this is more likely than not within 30 months" is enormous.
And notably: Andrej Karpathy just joined Anthropic specifically to build a team using Claude to accelerate pretraining research. That is recursive self-improvement in its early stage, being done deliberately. Clark's 60 percent estimate is not detached prophecy. It is Anthropic's internal view of the trajectory of work they are actively doing right now.
9. The AI Layoff Wave: Snap, Intuit, and Why This Pattern Is Different
Two companies announced AI-linked workforce reductions this week that follow a pattern now repeating across the technology industry.
Snap cut approximately 1,000 employees and closed 300 open positions in April, with the announcements continuing to ripple through the press this week. The company explicitly stated that AI generates more than 65 percent of its new code, and that it can operate with smaller teams because of AI agents handling work that previously required humans. Snap expects the restructuring to deliver over $500 million in annualized cost savings by the second half of 2026.
Intuit announced approximately 3,000 job cuts, roughly 8 percent of its total workforce, framing the reduction as freeing resources to invest in AI-driven product development across QuickBooks and TurboTax.
What is different about the 2026 wave of AI layoffs compared to previous technology downturns: companies are not cutting because business is slow. They are cutting because AI is doing the work that humans were doing, and the cost savings are immediate and measurable. Snap's stock rose on the announcement. Intuit's stock rose. The market is rewarding this trade.
This is the uncomfortable math of the AI productivity wave. Cost savings accrue to shareholders and AI infrastructure vendors immediately. Retraining and reemployment for workers happens on a much longer, much less certain timeline. Magnifica Humanitas published this same week is, in a very real sense, a direct response to exactly this dynamic.
The companies that follow Snap and Intuit this year will be numerous. Every organization with significant software engineering, customer support, or document-processing headcount is running the same calculation.
10. Anthropic Sues the Department of Defense
A significant legal story running in parallel to the funding and IPO news: Anthropic filed a lawsuit against the US Department of Defense, which designated the company a "supply chain risk" in March 2026 after Anthropic declined to allow its technology to be used for autonomous lethal weapons systems or mass surveillance of American citizens.
A federal judge issued a preliminary injunction blocking enforcement of the designation, meaning companies with DoD contracts can continue using Claude while the case is litigated. The case remains active.
Anthropic estimated the dispute put hundreds of millions to multiple billions of dollars of 2026 revenue at risk. The supply chain risk designation would have meant that companies doing business with the Pentagon would face restrictions on using Claude models, which is a significant category given that several major financial institutions in the Project Glasswing consortium also have government contracts.
This is the clearest public statement yet of Anthropic's hard lines: it will not build AI for autonomous weapons, will not allow Claude to be used to build them, and will litigate rather than comply with government pressure to change that position. OpenAI signed a DoD partnership for non-lethal military use. Microsoft is a major defense contractor. Google reversed its position on drone AI after Project Maven. Anthropic is the only frontier lab that has drawn a line and is fighting in court to defend it.
Pope Leo XIV's encyclical published this same week addresses autonomous weapons directly. Anthropic's co-founder Christopher Olah presenting at the Vatican while Anthropic litigates against the Pentagon over exactly this issue is not coincidental positioning. It is a coherent strategic statement about what kind of AI company Anthropic intends to be.
11. What Is Coming Next Week: WWDC, Microsoft Build, SpaceX IPO
The next two weeks are as event-dense as the past two. Here is what to watch:
Apple WWDC 2026 (June 8 to 12): The keynote is Monday June 8 at 10 AM PT. Expected: iOS 27 preview, Siri 2.0 with Gemini integration, the Extensions system for third-party AI in Siri, macOS 27 stability redesign, and potentially a preview of HomeOS for a tabletop smart home hub. Tim Cook is expected to helm the keynote. Bloomberg reports this could be his last as CEO before John Ternus takes over.
Microsoft Build 2026 (June 2 to 3): One week before WWDC. Fort Mason Center in San Francisco. Expected: major Azure AI Foundry updates, GitHub Copilot multi-agent orchestration announcements, a new AI Foundry for Windows SDK, and Satya Nadella's vision for Copilot as an agent-first multi-model platform. The conference is significantly condensed this year: two days, sharply focused on AI agents and enterprise developer trust.
SpaceX IPO pricing (June 11 to 12 expected): SpaceX filed its public S-1 on May 20 targeting a Nasdaq listing under ticker SPCX at a $1.75 trillion valuation. Pricing is expected around June 11 to 12, with listing in late June. This will be the first major data point for how the public market values AI-era infrastructure, and it will set the comparable for OpenAI's September listing.
I think WWDC is the most consequential of the three for consumers. The moment Apple shows Gemini-powered Siri responding in Dynamic Island on a live device, AI moves from something developers and enterprises use to something 2 billion iPhone users interact with daily. That is the distribution moment everything else has been building toward.
AI Weekly Scoreboard: Key Numbers From May 24 to 28, 2026
$900 billion: Anthropic's valuation after closing its $30B+ funding round, surpassing OpenAI's $852B private valuation.
$30 billion+: Amount raised by Anthropic in a single round, one of the largest private financing events in history.
42,300 words: Length of Magnifica Humanitas, Pope Leo XIV's AI encyclical.
245 paragraphs: Number of paragraphs in the encyclical, spread across five chapters.
1.4 billion: Catholics globally who will encounter Magnifica Humanitas, making it the widest-reach AI ethics document ever published.
1,000 jobs: Cut by Snap as AI generates 65%+ of its code, with restructuring delivering $500M+ in annualized savings.
3,000 jobs: Cut by Intuit in an AI-driven restructuring, approximately 8% of its workforce.
60 percent: Jack Clark's probability estimate that an AI system capable of training its own successor exists by end of 2028.
$1.25 billion per month: What Anthropic pays SpaceX for GPU compute access through May 2029.
$852 billion to $1 trillion: OpenAI's expected valuation range for its September 2026 public market listing.
Frequently Asked Questions
What AI news happened May 24 to 28, 2026?
The biggest stories were: Pope Leo XIV publishing Magnifica Humanitas, a 42,300-word AI encyclical addressing human dignity, labor rights, power concentration, and autonomous weapons; Anthropic closing a $30B+ funding round at a $900B valuation; China imposing overseas travel restrictions on top AI researchers at DeepSeek and Alibaba; Gemini Spark launching for Google AI Ultra subscribers; Apple creating genai.apple.com ahead of WWDC 2026 on June 8; and Snap and Intuit announcing AI-driven workforce reductions.
What does the Pope's AI encyclical Magnifica Humanitas say?
Magnifica Humanitas, published May 25, 2026, is Pope Leo XIV's first encyclical and the first Catholic document of this scale addressing artificial intelligence. Its 245 paragraphs across five chapters address: AI mimicking human identity and relationships in ways that undermine genuine human connection; AI-driven job displacement and the moral obligation to protect workers; the concentration of AI power among a few profit-driven companies; the ethics of autonomous weapons systems; and the risk that AI development without accountability could dehumanize society. The document calls for shared standards of social justice in AI development and urges governments and corporations to put human dignity above profit.
Why is Anthropic valued at $900 billion?
Anthropic's $900B valuation is supported by extraordinary revenue growth: the company projects $10.9 billion in Q2 2026 revenue, up 130% from Q1's $4.8 billion, with its first quarterly operating profit of approximately $559 million. The annualized revenue run rate is expected to surpass $50 billion by end of June 2026. Key revenue drivers include Claude Code (the dominant enterprise AI coding agent), Claude for Small Business (launched May 13, 2026), and enterprise contracts with PwC, Blackstone, Goldman Sachs, and others. Investors are also pricing in the expected October 2026 IPO and the improving compute cost ratio, which fell from 71 cents per revenue dollar in Q1 to a projected 56 cents in Q2.
Why is China restricting travel for AI researchers at DeepSeek and Alibaba?
China began imposing overseas travel restrictions on top AI professionals at private firms including DeepSeek and Alibaba in May 2026, requiring approval from government authorities before international travel. The restrictions are designed to prevent strategic AI knowledge from reaching foreign governments or companies through recruitment or legal processes. Chinese officials point to risks like the Meng Wanzhou case as precedents for why talent protection is necessary. The policy signals that China views its frontier AI researchers as national strategic assets, not ordinary private-sector employees.
What will Apple announce at WWDC 2026?
Apple's WWDC 2026 keynote is June 8, 2026, at 10 AM PT at Apple Park. Expected announcements include: Siri 2.0 with a chatbot-style redesign, Dynamic Island integration, conversation history, and multi-step task handling; the Extensions system for iOS 27, iPadOS 27, and macOS 27 that lets users route Siri to Claude, Gemini, and other third-party AI; Gemini integration under the hood via Apple's $1 billion per year licensing deal with Google; iOS 27 as a Snow Leopard-style stability release; macOS 27 with a Liquid Glass readability redesign; and potentially a HomeOS platform for a tabletop smart home hub.
What is Gemini Spark and how does it work?
Gemini Spark is Google's 24/7 personal AI agent launched at I/O 2026 and now live for Google AI Ultra subscribers at $100 per month. It runs on Google Cloud virtual machines even when your laptop is closed, working autonomously across Gmail, Google Sheets, Google Docs, Google Drive, and Calendar. It surfaces progress through Android Halo, a new notification layer in the Android status bar. Every action requires user approval before executing. MCP support for third-party apps like Canva, Instacart, and OpenTable is coming in weeks. Chrome integration follows this summer.
What is Jack Clark's intelligence explosion prediction?
Anthropic co-founder Jack Clark stated at the Oxford Cosmos Lecture on May 20, 2026, and in a subsequent Anthropic Institute research document, that there is a 60 percent or higher probability that by end of 2028, an AI system exists capable of training its own successor, a process called recursive self-improvement or "intelligence explosion." He also maintained a non-zero chance that AI could pose an existential risk to humanity. This is the first time a major AI lab has published a specific probability estimate and timeline for recursive self-improvement in an official document.
Why are so many companies laying off workers because of AI in 2026?
The 2026 AI layoff wave differs from prior technology downturns because companies are not cutting due to poor business performance. They are cutting because AI agents and models are handling work that previously required human headcount, with immediate and measurable cost savings. Snap stated that AI generates more than 65 percent of its new code and cut 1,000 jobs. Intuit cut 3,000 jobs while investing more in AI-powered QuickBooks and TurboTax workflows. Meta cut 8,000 jobs in late May while spending $125 billion on AI infrastructure. The pattern: AI productivity gains flow immediately to shareholders and AI infrastructure vendors; worker retraining happens on a longer, less certain timeline.
Recommended Reads on Unrot
What is an AI agent? A beginner's guide to autonomous AI in 2026
What is Apple Intelligence? iOS AI features explained for beginners
AI models explained: GPT-5.5, Claude, Gemini 3.5 compared for beginners
Daily AI learning beats trying to catch up weekly. Every time.
Five minutes a day compounds faster than you think. The best time to start was last month. The second best time is right now.
References
Vatican News: Pope Leo XIV's first encyclical Magnifica Humanitas to be published May 25
TIME: Pope Leo Uses First Major Papal Text to Warn About Dangers of AI
EWTN News: Magnifica Humanitas invokes justice to combat anti-human vision in AI
Bloomberg: Anthropic to Close Over $30 Billion Round as Soon as Next Week
Bloomberg: China Limits Overseas Travel for AI Talent at DeepSeek, Alibaba, Private Firms
Business Standard: Apple's gen AI website points to Siri overhaul ahead of WWDC 2026
MacRumors: WWDC 2026 Promises Apple Intelligence and Siri Upgrades
TechTimes: Anthropic Funding Round to Top $30B: $900B Valuation Would Surpass OpenAI
FOX Business: Snap cuts 1,000 jobs in AI-driven workforce restructuring