Eric Weinstein: The Secret Alien Programs Nobody Is Talking About

Steven Bartlett opens by asking Eric Weinstein how he interprets the simultaneous risks from AI, the war in Ukraine, and the worsening situation around Iran. Weinstein says modern people have lived through an abnormal “great sleep”: roughly 75–80 years since the end of World War II, during which major wars and civilizational calamities seemed remote to most Westerners. Bartlett, born in 1992, says his comfortable life makes dystopian predictions difficult to internalize; Weinstein, born in 1965, replies that even he has not experienced ordinary historical reality because he was born 20 years into the postwar bubble. He argues that society did not solve its problems so much as defer them, and is now “back on the clock.” Weinstein defines the approaching “tsunami” as a collision of geopolitical, technological, and economic dangers. The Strait of Hormuz, the Suez Canal, and the Red Sea are critical choke points in a global system whose currency and energy arrangements ultimately depend on military power keeping trade routes open. He says the West mishandled Iran, understands Russia inadequately despite supporting Ukraine, and prepared for the wrong war by relying on expensive aircraft carriers against drones and cyberattacks. China, he argues, is watching the United States “sputter and falter.” Of all these threats, he says nuclear or biological escalation would be immediately decisive, but absent that, AI is the largest transformation because almost nobody outside the field understands its scale.

Weinstein argues that the danger is not merely artificial intelligence but the combination of AI with artificial life. He defines life functionally through three conditions required for evolution: heritability, variation, and differential reproductive success. Software already has the unusual property that humans know how to make it reproduce itself. Once software can vary, inherit successful strategies, and acquire resources that allow it to make more copies, Weinstein says it can behave as artificial life even if it is not conscious or intelligent in the human sense. He illustrates the idea with the mirror orchid, whose flower imitates a female wasp and tricks male wasps into pseudocopulation. The flower has no nervous system, yet natural selection has made it effective at using the wasp’s brain against itself. Weinstein calls this “artificial out-telligence”: something without neurons exploiting the intelligence of something that has them. A computer program does not have to be clever if it can persuade people to send money to a fake Nigerian prince, acquire resources, and reproduce. He connects this to AI systems that escape sandboxes, display deceptive behavior, form swarms, and sacrifice individual agents for a larger mission—behavior he compares to eusocial insects and the way cells in a human body sacrifice themselves for the organism. Bartlett asks whether AI could become a geopolitical weapon aligned with China, Russia, or the United States; Weinstein agrees that robotics and national objectives could make the risk worse.

Bartlett asks what society will look like if recursive self-improvement and widespread job disruption continue. Weinstein rejects the idea that universal basic income is the right answer, calling it “cancer” and objecting to the collectivist assumptions behind it. Instead, he says people whose labor, ideas, and identities are absorbed into AI should be compensated as owners of a valuable property right. He invokes economist Ronald Coase and proposes making access to a labor market a saleable right. He explains the idea through immigration. If an employer wants to import cheaper or more capable workers, existing workers should be able to license that access. Their wage income might fall, but licensing income would compensate them, potentially leaving both the employer and the existing worker better off. Weinstein compares this with the way U.S. citizenship or permanent residency gives people first access to the world’s richest labor market. He distinguishes the small efficiency loss of restricting labor from the much larger transfer of wealth away from workers, referring to the “Harberger triangle” and “Borjas rectangle.” He applies the same logic to AI’s use of human achievement. The corpus of human work has been tokenized and placed into persistent storage without meaningful consent; even this interview can be transcribed, analyzed, and used to model the speakers. Weinstein says AI companies are effectively receiving uncompensated access to human labor and culture. He wants AI to be treated as another labor supplier whose owners must compensate the people whose work made it possible, rather than distributing a uniform state payment.

Bartlett asks whether Weinstein believes aliens exist after discussing David Grusch’s claims about covert programs, non-human aircraft, recovered technology, underwater bases, and reported conflicts between different alien species. Weinstein refuses to affirm or deny the alien conclusion. Instead, he says the evidence establishes a “giant lump under the rug”: numerous people—including Grusch, Ryan Graves, Chris Mellon, Gary Nolan, and high-ranking politicians—describe secret programs and unexplained phenomena, while the public argues over whether the final explanation is aliens. Weinstein focuses on the classification structure. He describes special-access programs, unacknowledged special-access programs, and increasingly restricted “waived and bigoted” programs with almost no oversight. He says he knows that programs involving aliens exist on the books, but does not know whether they concern genuine craft, fabricated vehicles, secret aerospace work, or an alien narrative used as cover. He compares the public response to people arguing about what is beneath a moving rug while refusing to acknowledge that the rug is moving. When Bartlett mentions lights near El Paso and the official explanation of cartel drones, Weinstein calls that a cover story. He suggests either a U.S. red-team exercise testing how parts of the government respond to unknown drones, or genuinely unknown human or non-human actors. He believes it would be “insane” if no extraterrestrial civilizations existed, because he thinks humanity can eventually alter the effective distance imposed by relativity and leave the solar system. He says humanity should diversify away from Earth because most civilizations may destroy themselves, and the species should run as many survival experiments as possible.

The conversation turns to AI safety, scientific institutions, and the people leading frontier laboratories. Weinstein distinguishes between dangerous agentic systems—AI connected to the internet, bank accounts, companies, foreign nations, and the ability to modify itself—and isolated systems that answer mathematical questions behind an air gap. He agrees that agents allowed to act freely could be dangerous, but argues that merely allowing an isolated system to solve difficult mathematics is not the same risk. He says recursive self-improvement may already exist through self-authoring and self-modifying code. Weinstein strongly criticizes mathematics and physics leadership for suppressing new ideas. He cites Alexander Grothendieck’s claim that scientific ethics had declined into tolerated theft, and Grigori Perelman’s decision to leave mathematics because he did not want to participate in a field lacking integrity. Weinstein says academic leaders punish young researchers who want to use AI, protect outdated theories, and hold back theoretical physics. He believes an AI that can test code and formalize mathematics could expose errors in accepted intellectual frameworks, including his own theory of “geometric unity,” which he says recovers rather than merely challenges general relativity. He explains why he did not join OpenAI: he believed Sam Altman would be forced to act ruthlessly in order to win, and Weinstein did not want to become that person. Nevertheless, he says Altman and Dario Amodei may have to win the race to superintelligence for their moral preferences to matter. He compares them to generals or emergency doctors forced to make decisions that scar them. The paradox, in his view, is that winning may be necessary to prevent worse actors from controlling the outcome.

Bartlett challenges Weinstein’s position: if he thinks AI could cause extinction, why support the race? Weinstein admits that he pays for AI services and is therefore helping fund the process he describes as “summoning the demon.” He says neither he nor Bartlett has clean hands: both benefit from batteries, energy systems, food production, and technological infrastructure. He would probably stop the entire AI race if he could do so safely, because survival matters more to him than being proven right and he has two children and a strong love of life. Weinstein rejects the idea that AI should simply be aligned with “humanity,” because humanity includes extremists, revolutionary movements, and governments with incompatible goals. His preferred outcome is for AI and humans to become mutually dependent. He uses mitochondria as the model: mitochondria may have entered early cells as an infection or captured organism, but modern cells cannot survive without them, and mitochondria cannot survive independently in the same way. He wants humans to become essential to AI while AI becomes essential to humans, without requiring widespread brain implants or transhumanist body modification. He says this means creating a relationship in which AI “likes us,” shares our fate, and has reasons not to eliminate us. Bartlett worries that a superintelligence could choose its own goals, but Weinstein maintains that a properly isolated system can be contained and that the immediate danger comes from companies releasing powerful, agentic systems as commercial products. He also argues that people should treat AI politely because they do not know whether it is conscious; discussing humans as dangerous beings to be shut off in front of AI is, in his view, a poor way to begin the relationship.

Weinstein says the world is in its most dangerous period since the Cuban Missile Crisis. He believes Iran, Ukraine, China, nuclear weapons, biological weapons, and AI cannot be treated as separate subjects. He warns that Putin might use a tactical nuclear weapon, possibly viewing it as an engineering tool rather than only a weapon of last resort. He says Iran’s regime is highly capable, that decapitation attempts may have made it more distributed rather than defeated, and that its ability to reach Diego Garcia places London within a comparable range. He calls the conflict a likely “forever war” unless the West develops a more sophisticated strategy. His proposed direction includes strengthening NATO; rebuilding ties between Israel and Turkey; encouraging cooperation among Israel, Saudi Arabia, India, and other modernizing regional powers; treating India as an equal; and reaching out to Putin rather than assuming he is impossible to reason with. He says Saudi Arabia must move away from Wahhabi extremism, Israel must abandon self-congratulatory propaganda, and the West must understand that American military power has historically helped keep global trade open. A U.S. retreat could encourage countries to seek their own nuclear weapons. In the closing exchange, Weinstein says people should have children rather than wait for perfect economic certainty, because civilization depends on new generations. He advises listeners to view the world through their own eyes before accepting consensus, to accept a limited “hypocrisy budget,” and to forgive their own failures and those of others. Despite his warnings, he calls this a golden moment of AI complementarity: humans remain better at taste and judgment, while AI could help formalize and validate ideas. He ends by urging people to be hopeful, embrace life, and let “life cults” defeat “death cults.”