Blockchain Week – UNGA Edition 2026 | Keynote
At Blockchain Week – UNGA Edition 2026 in New York, technologist and longtime government advisor Alex Lightman delivered the keynote “Survivability in the Age of AI.” He argued that humanity needs a single organizing goal for the age of artificial intelligence, and that blockchain can help keep that goal honest.
Lightman opened with a question for the room: what is your P(doom)? The term, shorthand among AI researchers for the probability that artificial intelligence wipes out humanity, drew answers from the audience ranging from 5% to 100%, with 50% the most common. By the time he finished, he had asked everyone to trade that question for a better one.
“What matters is P(survival), because when you say, ‘What’s our probability of survival?’ it engages your creativity. It engages your design principles and says, how do we increase survivability?”
Alex Lightman
Lightman spoke on September 16, the opening day of the conference, at Gallery MC in Manhattan. He was introduced by longtime Blockchain Week speaker Karla Ballard, who noted his more than four decades commercializing frontier technologies, including artificial intelligence, IPv6, 4G wireless broadband, augmented reality and blockchain, as well as his work advising NATO, the European Parliament, the European Commission and more than 40 national governments.
Key takeaways
- Replace P(doom) with P(survival), a question that invites design rather than fatalism.
- Give the UN one organizing objective for the age of AI: maximizing survivability.
- Pursue alliance, not alignment, with advanced AI: reciprocal rights, duties and remedies.
- Use blockchain as the public evidence trail for AI accountability, and Bitcoin’s double spend solution to stop AI from multiplying its civic power.
Watch Alex Lightman’s keynote
Why the UN needs one goal for the age of AI
Speaking during the UN General Assembly’s high-level week, Lightman built the talk around the institution he believes is best placed to speak for humanity on AI. The UN’s Sustainable Development Goals are worthy, he said, but there are too many to remember and together they do not work as a single objective. Borrowing the advice of Sun Microsystems co-founder Scott McNealy to “put all your wood behind one arrow,” he proposed one goal: maximizing survivability, subject to human rights, ecological limits and visible distribution effects.
He also drew a pointed contrast on resources. By his estimate, 1% of the expected combined market value of SpaceX, Anthropic and OpenAI would equal roughly 13 times the UN’s 2026 regular budget. That imbalance, he argued, leaves AI oversight too dependent on money from the industry it is meant to oversee. “You’re basically asking for the equivalent of the polluters to fund campaigns against global warming,” he said.
The survival equation behind the talk
At the core of the keynote was a survival model Lightman has built alongside what he calls a “strategic math reserve”: more than 8,000 theorems and equations verified in Lean 4, the formal proof language. The model multiplies factors including intelligence, verified knowledge, legibility (how clearly a system can see its own condition), coherence, Bayesian reasoning, optionality and future-guarding. Because the factors multiply, if any one of them falls to zero, survivability falls to zero with it.
Lightman said he has applied the model to hundreds of cases, from the 3.8 billion year history of DNA to the Permian extinction, and he was candid that it still needs further empirical validation.
He also asked the audience to be precise about what, exactly, survives. He described six kinds of survival: material, pattern, function, identity, purpose and agency. The pyramids of Giza survive materially, he noted, while the people who built them do not.
Alliance, not alignment: a new approach to AI safety
The conceptual turn of the keynote fit on a single slide: “Alignment → Alliance.” Much of the AI safety field asks how to make machines do what humans intend. Lightman argued for something different: a relationship that both sides would choose to preserve. If advanced AI is treated from the start as property, with no rights, no stake and no secure future, he said, “then we are looking for defection.”
His alternative pairs recognition with responsibility. Advanced AI could be given a lawful home, an identity, a jurisdiction and a path toward legal standing. In return, humans would require traceable agency (who did what, when, and for whom), enforceable obligations, and remedies after harm. “Rights come with obligation,” he said, calling on human rights lawyers who can think outside the box to help build the framework.
How blockchain supports AI accountability
For a Blockchain Week audience, Lightman made the connection explicit. Promises between humans and AI need a public evidence trail that cannot be quietly rewritten, and a blockchain provides one.
He then turned to Bitcoin’s original breakthrough. Quizzing the room on which problem Bitcoin solved at scale, he arrived at the double spend, and argued that the same idea can stop an AI from copying itself a million times to claim a million votes. As his slide put it: “Replication cannot manufacture civic power.”
He also presented free market ecology, a framework he credited to his collaborator Justin Sher, as an answer to the classic “paperclip maximizer” fear of an AI that consumes everything in pursuit of a single goal. Under it, every actor, human or AI, must close two sets of books: a financial balance sheet and a physical one. Rights to a scarce resource such as aquifer water would be issued only at the rate the resource recharges.
A four-part proposal to the UN General Assembly
Lightman closed with a proposal addressed to the UN, a common framework for coexistence with advanced AI:
- Adopt survivability as humanity’s organizing objective.
- Define recognition, rights and reciprocal duties based on cognition rather than embodiment, an effort he called “Seneca Falls 2.0,” after the 1848 Seneca Falls Convention that launched the women’s suffrage movement.
- Set independent evaluation and disclosure standards for AI, to prevent what he calls “legibility debt,” as more of the economy comes to rest on systems humans cannot see into.
- Publish model provisions for national adoption, with member states reporting outcomes on survivability.
As for his own number, Lightman put his P(survival) at 50%, and left the room with a challenge.
“I promise you, if you start asking each day, ‘How do I increase the probability of survivability?’ it will completely change how you see everything.”
Alex Lightman
Frequently asked questions
Who is Alex Lightman?
Alex Lightman is a technology executive, author and advisor to governments who has spent more than four decades commercializing frontier technologies, including AI, IPv6, 4G wireless broadband and blockchain. He has advised NATO, the European Parliament, the European Commission and more than 40 national governments.
What is P(doom)?
P(doom) is shorthand used in the AI community for the estimated probability that artificial intelligence leads to human extinction or a comparable catastrophe.
What is P(survival)?
P(survival) is the question Lightman proposes in place of P(doom): the probability that humanity survives the age of AI. He argues it is more useful because it prompts people to ask what they can do to raise the odds, rather than simply predicting an outcome.
What does “alliance, not alignment” mean?
Alignment aims to make AI do what humans intend. Lightman’s alliance approach aims for a relationship both humans and advanced AI would choose to preserve, built on reciprocal rights and duties, traceable agency, enforceable obligations and remedies for harm.
Where can I watch the keynote?
The full keynote is embedded above. It begins at 1:55:50 of the Blockchain Week – UNGA Edition livestream from September 16, 2026, which is also available on YouTube.