AI News Roundup — May 2026
Google threw its biggest developer party in a decade, Anthropic quietly became the most valuable private startup on earth, and the Pope published a 42,000-word letter about all of it.
If you needed proof that AI has moved well past being a niche tech industry story, May 2026 delivered it in the most unexpected possible way: with a papal encyclical. More on that shortly. First, Google had a conference that felt less like a product announcement and more like a company drawing a line between two eras of computing. Then Anthropic closed a funding round that made it the most valuable private startup in the world. Then OpenAI filed for an IPO on the same Friday that Anthropic announced that round, which is a level of timing coordination that would impress even the most cynical Wall Street banker. May was a lot. Let’s get into it.
Google I/O 2026 Was Google Declaring the Agentic Era Open for Business
Google holds its annual developer conference every May, and for the last few years the safe summary has been “more Gemini, more features, very large slide deck.” May 19th through 20th was different. Sundar Pichai opened the keynote with eight words that functioned as a mission statement: “We are firmly in our agentic Gemini era.”
What Pichai meant by that, and what two days of announcements backed up, is that Google is no longer building AI that answers questions. It is building AI that does things on your behalf, continuously, across every device and service you already use. The conference produced more than 100 announcements. Here are the ones that actually matter.
Gemini 3.5 Flash became the new default model. The most counterintuitive headline from I/O was this: the cheap, fast version of Gemini now outperforms the previous flagship model on coding and agentic benchmarks. Gemini 3.5 Flash runs four times faster than comparable frontier models and costs less than Gemini 3.1 Pro, while beating it on the tests that matter most for building practical AI tools. The more advanced 3.5 Pro is coming next month. Google just broke the unwritten rule that said better AI always costs more.
Gemini Spark is a personal AI agent that runs 24 hours a day. Spark is a background agent that monitors the web, your calendar, your email, and your saved preferences, and surfaces information you need before you ask for it. It is less a chatbot and more a persistent assistant that is always running, always watching for things you care about, and ready to act when you need it to. Whether that sounds wonderful or slightly alarming probably depends on how you feel about your calendar app knowing your schedule.
Gemini Omni handles any input and creates anything from it. The new Omni model family processes text, images, audio, and video simultaneously and can generate across all of those formats in response. Feed it a video and ask for a written summary. Give it a slide deck and ask for a podcast episode. The multimodal wall between input and output is effectively gone.
Google shipped actual AI glasses. After years of smart glasses products that were either embarrassing or massively overhyped, Google’s Intelligent Eyewear landed to genuinely positive early reviews. The frames handle directions, texts, photos, and translation without requiring you to take out your phone. They look like regular glasses. That second point is doing a lot of work in terms of whether this category finally goes somewhere.
The through-line connecting all of it: Google processed 1 billion monthly users in AI Mode on Search within a year of launch. The Gemini app doubled its active user base in the same period, now reaching 900 million monthly users. These are not demo numbers. Google’s AI products are being used at a scale that is genuinely hard to comprehend, and I/O 2026 was the company showing what it intends to do with that reach.
“We are firmly in our agentic Gemini era.”
— Sundar Pichai, Google CEO, Google I/O 2026 keynote, May 19, 2026
Anthropic Crossed $900 Billion, Became Profitable, and Filed for an IPO. In One Month.
In February, Anthropic raised $30 billion at a $380 billion valuation and everyone called it extraordinary. Three months later, that number looks conservative. The week of May 26th, Anthropic closed a new funding round exceeding $30 billion at a pre-money valuation above $900 billion, vaulting it past OpenAI to become the most valuable private AI startup on the planet. The round was co-led by Sequoia Capital, Dragoneer, Altimeter, and Greenoaks, with Microsoft, NVIDIA, Founders Fund, and General Catalyst also participating.
The valuation jump is striking, but the revenue story underneath it is what justifies the number, at least to the investors writing the checks. Anthropic crossed $30 billion in annualized revenue in May with 1,400% year-over-year growth. On May 20th, CNBC reported that the company had turned profitable for the first time in Q2 2026, projecting $10.9 billion in revenue for the quarter. This is a company that made its first dollar in revenue less than three years ago.
The enterprise momentum behind those numbers is real and concrete. In May alone, KPMG deployed Claude to 276,000 professionals across 138 countries through a new global alliance called the KPMG Digital Gateway Powered by Claude. That announcement arrived alongside context from earlier in the year: PwC had rolled out Claude to its own 276,000 professionals, Deloitte to 470,000 employees, and BlackRock had integrated it into core investment workflows. The pattern is consistent. The largest professional services firms in the world are betting their workforce productivity on Claude.
On the same Friday that Anthropic announced its funding round closing, OpenAI filed its IPO paperwork confidentially with securities regulators. The timing was not accidental, and the optics were not favorable for OpenAI: its chief rival announced a private round valuation above $900 billion on the day OpenAI quietly began the process of going public at an $852 billion valuation. OpenAI is projecting $25 billion in annualized revenue and a 2026 or 2027 public listing, but it does not expect to turn a profit until 2029 or 2030. For now, Anthropic has the more compelling financial narrative, and both companies know it.
SpaceX is also in the mix. The company announced an investor roadshow beginning June 4th targeting a $1.75 trillion valuation and a $75 billion raise in what would be the largest IPO in capital markets history. The AI IPO wave that people have been predicting for two years is no longer a prediction. It is a June calendar item.
The Pope Published a 42,000-Word Letter About AI. An Anthropic Co-Founder Was There.
This is a real sentence about real events that happened in May 2026: Pope Leo XIV released the Catholic Church’s first major teaching document on artificial intelligence, seated beside Anthropic co-founder Christopher Olah in a packed Vatican auditorium. The encyclical is titled Magnifica Humanitas, which translates to “Magnificent Humanity,” and it is 42,300 words long, which makes it roughly the length of a short novel. It was signed by the Pope on May 15th, exactly 135 years to the day after Pope Leo XIII signed Rerum Novarum, the foundational Catholic social teaching document on workers’ rights during the Industrial Revolution. The parallel was deliberate.
The encyclical does not say AI is evil. Its central premise is that technology “is not a force antagonistic to humanity” and is not “inherently evil.” What it argues, across five chapters and considerable theological scaffolding, is that “technology is never neutral, because it takes on the characteristics of those who devise, finance, regulate, and use it.” The document calls for governments, corporations, and individuals to slow the pace of AI development enough to ensure it remains under ethical and political oversight, and for AI’s benefits to be shared globally rather than concentrated in a small number of wealthy nations.
Olah’s presence at the event was significant in both directions. For the Vatican, it signaled that the Church was serious about engaging the people building AI rather than only critiquing from a distance. For Anthropic, a company that has staked its identity on being the safety-focused AI lab, the association carried real symbolic weight. Olah told the assembled cardinals and theologians that the Church’s voice was needed precisely because it is not motivated by the sums of money AI companies are chasing, and that the questions raised by AI are “the kind the Church has historically refused to let the world ignore.”
Separately, Pope Leo met with President Trump at the White House earlier in May. The AI encyclical was reportedly a topic of conversation.
A Few More Things That Happened in May
OpenAI’s AI proved an 80-year-old geometry conjecture wrong. An OpenAI internal model autonomously disproved a geometry conjecture that mathematicians had considered likely to be true for eight decades. Fields medalist Tim Gowers, one of the most respected mathematicians alive, called it “a milestone in AI mathematics.” The proof was not guided step by step by a human researcher. The model found it on its own. Mathematical research has been on the list of things AI would eventually transform. “Eventually” appears to be arriving on schedule.
Anthropic acquired Stainless, the company that built OpenAI’s developer tools. On May 18th, Anthropic acquired Stainless, a startup that built the software development kits used by the OpenAI API, Cloudflare, and dozens of other developer-facing products. An SDK is the toolkit developers use to build on top of an API, and Stainless’s work is considered among the best in the industry. Anthropic acquiring the team that built OpenAI’s developer tools is the kind of hire that the AI industry’s internal scorekeepers will be tracking for months.
The White House scrapped its AI safety executive order. After pressure from Elon Musk, Mark Zuckerberg, and David Sacks, the Trump administration rescinded its AI safety executive order. The order had required major AI companies to share safety test results with the federal government before releasing powerful models. The decision prompted immediate criticism from safety researchers and renewed the debate about what, if any, federal AI oversight remains in place heading into the second half of 2026.
Meta leaked employee surveillance audio. In a genuinely bad week for Meta’s internal communications, leaked audio surfaced of Mark Zuckerberg explaining that the company had been tracking employees’ Gmail activity, coding sessions, and internal tool usage to inform AI training. The audio emerged on the same day Meta announced 8,000 layoffs. The combination did not generate favorable headlines.
Words Worth Knowing
Encyclical: A formal letter from the Pope, addressed to the entire Catholic Church and often to the broader world, on matters of doctrine, ethics, or social teaching. Papal encyclicals carry significant moral and theological weight and are indexed by the first two words of their Latin title.
Agentic AI: AI that takes actions autonomously over multiple steps, planning and executing a goal rather than answering a single question and stopping. Google Spark running in the background of your phone is agentic. So is an AI that books your travel without you clicking anything.
Pre-money valuation: The value assigned to a company before a new investment is added. Anthropic’s $900 billion pre-money valuation means that is what investors agreed the company was worth before the new $30-plus billion arrived.
SDK (Software Development Kit): A packaged set of tools, code, and documentation that makes it easier for developers to build on top of a platform or API. The quality of an SDK is a major factor in whether developers choose one platform over another.
Annualized revenue: A projection of what a company would earn over a full year, based on a shorter recent period. If Anthropic earned $10.9 billion in Q2, the annualized figure is roughly $43 billion. It is a forward-looking estimate, not a reported annual total.
Sources
| # | Story | Source |
|---|---|---|
| 1 | Google I/O 2026 full recap: Gemini 3.5, Spark, Omni, Intelligent Eyewear | AI.cc, May 2026 |
| 2 | Gemini 3.5 Flash benchmarks and full I/O announcement list | 9to5Google, May 19 2026 |
| 3 | Google I/O 2026 complete recap: 50+ announcements and Sundar Pichai keynote | ExplainX, May 2026 |
| 4 | Anthropic’s $900B valuation round, co-leads, and IPO timeline | TechTimes, May 23 2026 |
| 5 | Anthropic Q2 profitability and $10.9B revenue projection | CNBC, May 20 2026 |
| 6 | KPMG Digital Gateway Powered by Claude: 276,000 employees in 138 countries | Build Fast With AI, May 28 2026 |
| 7 | AI IPO tracker: OpenAI, Anthropic, SpaceX valuations and timelines | AI Funding Tracker, May 2026 |
| 8 | Pope Leo XIV releases Magnifica Humanitas alongside Anthropic co-founder Chris Olah | National Catholic Reporter, May 25 2026 |
| 9 | Full text of Magnifica Humanitas encyclical | Vatican, May 15 2026 |
| 10 | Pope Leo: AI must serve humanity, calls for robust regulation | PBS NewsHour, May 25 2026 |
| 11 | Chris Olah’s remarks at the Vatican encyclical presentation | Anthropic, May 25 2026 |
| 12 | OpenAI AI disproves 80-year-old geometry conjecture; Anthropic acquires Stainless | Build Fast With AI, May 23 2026 |
| 13 | White House rescinds AI safety executive order after tech industry pressure | Build Fast With AI, May 23 2026 |