Singularity University

10/27/2026

AI is changing the cybersecurity landscape, and the PHP2Shell vulnerability shows how fast the stakes are rising.

PHP is a software that powers up to 70% of the world’s websites. Researchers used AI to uncover a critical flaw in PHP’s own memory handling. The vulnerability could let attackers run code remotely without logging in, or read sensitive information like passwords, authentication tokens, and API keys from server memory.

But AI is accelerating both sides of the equation. The same capabilities that help researchers find vulnerabilities like this one can help hackers take advantage of security flaws. Meanwhile, the growing volume of AI-assisted discoveries is overwhelming open-source projects with bugs that need to be investigated and fixed.

That means there’s a shrinking window between discovering a vulnerability and seeing it exploited. As AI helps researchers find more bugs faster, keeping software secure will require faster patching, better coordination, and the resources to keep up with the growing workload.

10/23/2026

Congratulations to Ignacio Juarez, Singularity Fellow for Energy, on the progress at EnnoFlow.

Ignacio started his career in energy and saw firsthand how much time, capital, and physical resource it takes to build new capacity. As AI demand grows, energy, data, and compute are becoming more tightly linked, which led him to ask how critical infrastructure could expand at the pace of that demand. His team built its own answer in hardware and software. EnnoBox turns a connected location into a compute node, and Enna coordinates those nodes as a flexible mesh, shifting workloads in real time based on the energy and resources available. AI capacity can then grow node by node using space, power, and connectivity that already exist.

Ignacio brings this builder’s perspective to his sessions with Singularity, where he talks about what it takes to move from technological possibility to systems people and businesses can actually use. His vision opens critical infrastructure to participation from homes, small businesses, and large enterprises alike. We’re proud to support his work.

Follow EnnoFlow Technologies for more on their technology, upcoming deployments, and partner announcements.
https://hubs.li/Q04CcZvG0

AI demand is everywhere. Infrastructure should be too.

10/14/2026

If you haven’t finished your 2027 planning yet, don’t stress. The world is scheduled to end next month.

Okay, probably not. But this particular doomsday involves a singularity, so of course we have to weigh in.

In 1960, physicist Heinz von Foerster and his co-authors published a paper in Science predicting that humanity would be squeezed out of existence on November 13, 2026. Their equation had the global population accelerating so fast that it would hit infinity on that date, give or take five years. Mathematicians call that kind of point on a curve a singularity.

Von Foerster was building on an older fear. Thomas Malthus worried in the late 1700s that a growing population would outrun its food supply, and Paul Ehrlich revived that worry with The Population Bomb eight years after the Doomsday paper. Von Foerster believed human ingenuity would keep us fed. What he couldn’t picture was ingenuity changing the shape of the curve itself.

But it did. The global population growth rate peaked in the early 1960s, right around the time the paper came out. The boom came from germ theory, sanitation, and better nutrition keeping more kids alive. The Green Revolution let farmers grow far more food, and contraception and education gave women more say in how many children they had. Today, about 55 percent of countries have fertility rates below replacement level. Those shifts came from technology and knowledge that von Foerster’s equation couldn’t account for.

Our cofounder Peter Diamandis has been making this case for years, and he made it again on Fox Business this week. In his book Abundance, he calls technology “a resource-liberating mechanism” that takes a scarce resource and makes it abundant.
Doomsday forecasts tend to make the same mistake. They take today’s tools, project today’s limits forward, and assume the toolkit will stay frozen. We won’t be solving tomorrow’s problems with today’s technology, and forecasts that forget this miss the mark.

Abundance still takes work, though. Around 22 percent of children under 5 worldwide still experience stunted growth from malnutrition, and that number won’t drop on its own. Making something abundant and getting it to the people who need it are two different jobs.

We get to decide what tomorrow’s technology looks like. Every tool that bent the population curve came from people who chose a problem and refused to accept it as permanent. Von Foerster’s equation treated humanity like an experiment running on its own inside a sealed box. We have far more say over our circumstances than that, and we already hold the tools to make the future abundant. The question is: how will we step up to do it?

10/12/2026

A Stanford team spent years building a robot that flies like a bird, without the vertical stabilizers conventional aircraft require. They studied pigeon wing mechanics, modeled the aerodynamics, engineered synthetic alternatives.

Then they discovered they couldn’t manufacture feathers that matched the mechanical properties of real ones. Their advanced robotic bird flies with actual pigeon feathers.

Four billion years is a considerable head start.

That detail points to something larger. We tend to think of biomimicry as observing nature, then recreating it in human-made materials. But synthetic biology is producing a different possibility entirely: biology stops being a source of inspiration and becomes part of the technology itself. A plant that naturally accumulates metals from soil doesn’t just inspire a mining technique — it could become the mining process. DNA doesn’t just suggest how to store information more densely — it could become the storage medium.

The boundary between studying biology and building with it is becoming harder to locate. What that opens up across industries that don’t currently think of themselves as biotech is the more interesting story.

Read the full story, link in comments.

10/09/2026

This year’s Nobel Prize in Chemistry honors Henri B. Kagan and Kenso Soai, two chemists who recreated in the lab something only life itself had achieved.

Many molecules come in mirror-image forms, a little like left and right hands, yet living cells use only one version. For over a century, scientists couldn’t reproduce that one-handedness in a lab, where reactions produced an even mix of both. In the early 1980s, Kagan used catalysts to tip the balance toward one form, and Soai built on that work to design a reaction that produced only one.

Their discoveries let chemists choose exactly which version of a molecule to make, which shapes how we manufacture drugs, flavors, and fragrances. They also offer a possible explanation for how life on Earth settled on one handedness in the first place.

At Singularity, we believe biotech is approaching the same exponential curve AI has been riding.

Congratulations to Henri B. Kagan and Kenso Soai on this well-deserved recognition.

10/08/2026

Slime mold has no brain, yet a model of its behavior designed a subway network for Toronto that outperformed the real one on resilience, efficiency, and travel time.

The organism works without a central plan. Each tendril responds to what’s directly around it, and those local decisions add up to a network that keeps running when one part fails. Study author Raphael Kay noted that slime mold has had hundreds of millions of years of evolution to practice these problems.

Supply chains, logistics routes, and distributed teams face the same challenge: moving resources through a network that has to bend without breaking. Leaders who only study human-made models are working from a much smaller set of tested designs.

We explore this in our new book, Living Tech: The Convergence of Biology and Innovation.

Slime mold could map the fastest picking routes through a warehouse, the most resilient delivery network, or the smoothest flow of people through a hospital. Where in your business could a slime mold experiment uncover a better design?

10/07/2026

What does AI sovereignty actually mean for a country, a company, and for you? Does the old assumption, ’control the land and you control your fate,’ still work once the thing to control has no body and no border?

Dr. David Bray has spent his career considering these implications. He led the technology response for the CDC's Bioterrorism Preparedness and Response Program during 9/11 and the anthrax attacks, spent time on the ground in Afghanistan in 2009, later served the Senior National Intelligence Service Executives as executive director for a bipartisan National Commission for the Review of the R&D Efforts of the U.S. Intelligence Community, and served as Senior Executive and CIO with the FCC for four years where he and the team that he led won the Global CIO 100 award twice. Today, he's Principal at LeadDoAdapt Ventures and Chair of the Accelerator at the Loomis Council at the Stimson Center, advising governments and Fortune 500 leaders on AI and space.

We'll be addressing these questions and more with him next week on The Discussion Series.

10/06/2026

What happens when you spend 5 days in Silicon Valley rewiring how you think about the future and your role in it? Hear directly from our alumni on the impact of the Singularity Executive Program.

Our next cohort runs October 25 to 29 at the Ameswell in Mountain View, California. Don't miss out. Join the waitlist or apply for May 2027 at http://su.org/executive-program

10/02/2026

September was one of the busiest months of the year for AI model releases, with several major models dropping in 30 days and open-weight models gaining ground on the frontier.

Anthropic opened the month with Claude Fable 5.1, built for coding and knowledge work. Google and Meta followed with Gemini 3.8 Flash and an updated Muse Spark 1.3, respectively, which both target coding and agentic workflows. OpenAI released GPT-6 Astra, which it calls its most capable model yet, with gains in computer use, coding, cybersecurity, and scientific work.

And that was just the beginning. Anthropic released Opus 5.5 and Sonnet 5.5. OpenAI followed Astra with GPT-6 Sol and Luna, bringing flagship capabilities into faster, cheaper territory. xAI’s Grok 4.7 brought gains in reasoning and agentic tasks.

Meanwhile, open-weight models keep coming down the pipeline—and they’re only getting better. Xiaomi’s MiMo-V2.6-Pro got one of the highest scores to date for an open-weight model on the Artificial Analysis Intelligence Index. DeepSeek dropped V4.1 Flash, a smaller, faster model built for higher throughput and lower inference costs. Alibaba released Qwen-Image-2.1, which edged out Google's Nano Banana 2 on Qwen's own benchmark.

September's releases show how quickly the AI landscape is expanding in multiple directions at once: as frontier labs move the needle on capability, open-weight models are making advanced AI more accessible and deployable.

What will October bring?

09/30/2026

AI is often described as if it has a momentum of its own — advancing faster than we can understand it, producing outcomes no one intended, and forcing leaders to choose between accelerating and hitting pause.

But what if that framing is missing the most important part of the story? Behind every AI system are choices: what it’s optimized for, how it’s trained, which tradeoffs are accepted, and whose interests those decisions serve. Understanding those choices is becoming essential for leaders trying to separate what AI can actually do from what the dominant narrative says it can do.

In this conversation, Singularity Expert and VP of Research Alix Rübsaam will unpack the assumptions shaping today’s AI debate, explore where human agency and accountability still sit within increasingly complex systems, and challenge the idea that today’s approach to AI is the answer to every problem. Using real-world examples and current AI headlines, Alix will offer a practical lens for asking better questions about capability, safety, control, and the decisions behind the technology.

You’ll come away better equipped to look beyond the headlines, identify the human choices embedded in AI systems, and make more informed decisions about where — and how — AI belongs in your organization.

Link to full discussion in the comments.