(RNS) — Last month, a man hurried up the driveway of OpenAI CEO Sam Altman’s San Francisco compound hours before sunrise and lobbed a flaming Molotov cocktail at the gate. 

No one was injured, and shortly after, a suspect was arrested outside the offices of OpenAI, the company that makes ChatGPT, allegedly threatening to “kill everyone inside.” Daniel Moreno-Gama, 20, who has no prior criminal record, has been charged with 13 felonies, including attempted murder.

Moreno-Gama’s parents and lawyer have said he has a history of mental illness, but his digital footprint also shows he was deeply concerned that artificial intelligence would lead to human extinction. He frequented an anti-AI Discord server, where he mused in a chat about violent acts against tech leaders like Altman. And in a podcast interview recorded in January, he described himself as an “AI safetyist” and cited the influence of Eliezer Yudkowsky, a prominent AI theorist who argues that if anyone builds superintelligent AI, everyone will die. 

The attack on Altman has ratcheted up existing tensions in Silicon Valley. AI safety proponents overwhelmingly denounced Moreno-Gama’s alleged actions, including Yudkowsky, who insisted that “only law can prevent extinction.” Yudkowsky also assigned some blame to another camp: accelerationists, who largely believe the existential risk of AI is low and that unregulated AI development is needed to bring about a glorious utopia. 

In turn, accelerationists accused those known as doomers, like Yudkowsky, of radicalizing young people. “Proselytizers of an apocalypse cult bear moral and legal responsibility for violence committed by their followers,” venture capitalist and techno-optimist Marc Andreessen wrote on X the same day as the attack, accompanying the tweet with a photo of Charles Manson.

https://t.co/qWSdGA8O7h pic.twitter.com/Nd5n6Dqz5o

— Marc Andreessen 🇺🇸 (@pmarca) April 10, 2026

For doomers and accelerationists, ideological differences about AI can seem like paradise versus apocalypse. But in the view of philosopher Émile P. Torres, an intellectual historian and affiliate scholar at Data & Society, an organization focused on the social implications of emerging tech, these divisions are ultimately familial. 

“Family disputes can be vicious, but they are among family members who share most everything else,” Torres, who has studied these communities for years, told RNS. 

In 2023, Torres and computer scientist Timnit Gebru introduced the acronym TESCREAL, contending that an overlapping bundle of futuristic beliefs — transhumanism, extropianism, singularitarianism, cosmism, rationalism, effective altruism and longtermism — is propelling the race for artificial general intelligence. And while critics, especially those in the bundle, see their framework as too broad or slanted to be useful, Torres argues that TESCREAL is effectively a “secular religion,” big enough to encompass a “clash of eschatologies” and having shaped all the major AI companies today. 

‘A religion-shaped hole’

For Torres, who recently completed a postdoctoral fellowship at Case Western Reserve University in Cleveland, this research is personal. After walking away from the evangelical faith they grew up in, they embraced transhumanism, a movement focused on transcending the limits of the body, mind and lifespan through technology. It filled “a religion-shaped hole,” Torres has acknowledged. 

In the early 2010s, they wrote about AI and the future of civilization alongside some of transhumanism’s biggest names. Now, they consider themself an “ardent apostate” from the TESCREAL movement.

---

---

What unites those in TESCREAL is a fervent faith in AI’s salvific potential, helping humanity reengineer itself and colonize the universe — despite disagreements over how to go about it — Torres explained. They see plenty of similarities to Christian theology, too. 

“TESCREALism promises that you can gain immortality, you will eventually live in paradise where suffering has been abolished, and you can create a heaven in the literal heavens,” they said.

Torres also sees reverberations of the Rapture in the singularity — the theorized moment when AI surpasses human intelligence — and echoes of final judgment in visions of AI apocalypse. “There’s even a possibility of resurrection: If you don’t live long enough to live forever, you can have your body cryogenically frozen,” they said. 

Greg Epstein, a humanist chaplain and affiliate at the Berkman Klein Center for Internet & Society at Harvard University, covered TESCREAL ideologies in his recent book, “Tech Agnostic,” and also found religious parallels. 

“It’s not only incredibly religious — it’s much more of an influential religion than most of the religions in the world today,” he said.

Unpacking TESCREAL

The TESCREAL bundle is roughly chronological, starting with transhumanism. The term was first popularized in 1957 by Julian Huxley, an evolutionary biologist who was also an outspoken eugenicist. 

The underlying “racist, xenophobic, ableist, classist and sexist attitudes” of eugenics are shot through TESCREAL, Torres and Gebru argued in their 2023 paper. 

According to Torres, transhumanism is the core of TESCREAL. It’s present in the next ideology, extropianism — a 1990s-era community that earnestly discussed then-radical ideas like AI, mind uploading and multiplanetary civilization. It informs singularitarianism, the belief that the “singularity is a good thing, and that we should help make it happen,” as Yudkowsky wrote in 2000. And it’s evident in cosmism, which presents a future in which humans “merge with technology,” fulfilling “most of the promises of religions” and beyond, as computer scientist Ben Goertzel envisioned.

Rationalism, the next ideology, coalesced around LessWrong, a web forum founded by Yudkowsky in 2009, focused on honing reasoning skills to confront big ideas. Effective altruism, or EA, arose in tandem, applying a similarly rational approach to ethics and charity. 

Some of these aims may sound disconnected from, say, trillion-year timescales or digital deities, but Torres and other scholars argue they’re not. Long before AI was a mainstream topic, many rationalists and EAs debated its risks and embraced its supernatural-sounding possibilities.

Out of EA grew longtermism: an ethical outlook that treats “positively influencing the long-term future as a key moral priority of our time,” as philosopher and EA co-founder William MacAskill defines it. The concept has received plenty of attention and praise, including from Elon Musk. But detractors say that extreme longtermism can treat future humans — even post-human “digital beings” — as more valuable than those living now. 

In a forthcoming entry in the Oxford Research Encyclopedia on TESCREAL, Torres draws links between the bundle and all the frontier AI companies. 

Altman, for instance, has credited Yudkowsky as the person who “got many of us interested in AGI” and who was “critical in the decision” to start OpenAI. The CEO has also invested tens of millions into longevity research and expressed belief in a cosmism-style “merge,” suggesting that “we will be the first species ever to design our own descendants.” 

As for OpenAI’s rival Anthropic, which makes the Claude chatbot, “it’s essentially an EA company,” Torres said.

Anthropic was founded by former OpenAI researchers, many with EA ties, who left due to concerns that Altman’s company was not sufficiently safety-focused. They started Anthropic with funding from EA-associated backers, including Holden Karnofsky, who is co-founder of the EA nonprofit Open Philanthropy and is married to Daniela Amodei, Anthropic’s president. Karnofsky also joined Anthropic last year.

Over the last three years, TESCREAL has been widely covered and cited, including in a new documentary, “Ghost in the Machine.”

Not all scholars are sold on it, however. 

Seth Lazar, a philosopher of machine intelligence at the Australian National University, wrote in an email that TESCREAL “is highly polemical, and ignores many other strands in the history of AI that do not meet that narrative.” If there has been little “serious intellectual engagement” with it, that’s because Gebru and Torres consistently use “social media to attack anyone who disagrees with them,” Lazar said. 

Torres pushed back, noting that TESCREAL is “relatively new” and being employed with “growing frequency.”

“As for Timnit’s and my approach being polemical, I think that’s in the eye of the beholder,” Torres said. Given how “outrageous, dangerous, often racist, yet powerful the TESCREAL movement is … I’d argue there’s good reason to be a bit upset about that.”

‘Shut it down’ vs. ‘Let it rip’ 

When Torres and Gebru first introduced the TESCREAL acronym in 2023, a philosophy called effective accelerationism was gaining traction. E/acc, as it’s known, champions rapid, unregulated AI development. It was meant to counter — and even mock — EA’s safety-first ethos.  

As figures like Yudkowsky have increasingly implored Silicon Valley to “shut it down” as AI advances rapidly, e/acc’s rallying cry became “let it rip.”

“We have no affinity for biological humans or even the human mind structure,” declared an early e/acc website, pledging to ramp up AI development “to preserve the light of technocapital.” Not accelerating, e/acc proponents like Guillaume Verdon say, is the real existential threat.   

E/acc didn’t officially make the TESCREAL acronym, but Torres sees it as part of the bundle anyway. EAs and e/accs differ on timing, they explained — many EAs don’t think society is ready for superintelligence in the near future, while e/accs think we are. Otherwise, their ultimate goal is identical, Torres argues: “to go out, colonize space, and to build a sprawling, multigalactic civilization.”

Epstein likens these ideological differences to “internecine skirmishes” between “sects or denominations,” where each side is so “deeply confident that their view of AI justifies almost anything,” he said. “They really want to convert us all.”

---

---

For their part, Torres believes that any attempt to realize TESCREAL’s techno-utopian “mirage” will result in human extinction. “Even the best case outcome with the AGI race is one in which our species is sidelined, marginalized, disempowered and eliminated,” they said. 

The divisions within Silicon Valley circles are stark, too. It’s one reason Ozy Brennan, a rationalist writer and activist, has critiqued the “TESCREAL bungle,” as they put it. To them, it feels dishonest to lump techno-optimist Andreessen in with their own community. 

Andreessen expects AI “to be transformative the same way the internet was,” Brennan argued. Conversely, many AI safety advocates believe AI will “be able to kill everyone or to build a Dyson Sphere out of the asteroid belt in which we all live as immortal superbeings without material scarcity.”

“That seems like an important disagreement to me,” Brennan said. 

Brennan also takes issue with Torres’ insistence that transhumanists like Yudkowsky are pro-extinction. “Eliezer Yudkowsky is one of the most profoundly anti-death people I have ever met,” they said. “Torres is allowed to believe that a transhuman or upload with my memories, personality and preferences is ‘not Ozy.’… But I find their level of vitriol and condescension to be really unenlightening.”

Still, Brennan acknowledges some overlap between “belief systems” on the spectrum between e/accs and EAs. “The thing Lutherans hate the most is a different kind of Lutheran,” they said. “A lot of rationalists and effective altruists end up disliking Marc Andreessen (because) we have very similar views in a lot of ways.”

As someone deeply enmeshed in the Bay Area’s rationalist and EA scene, Brennan has grown accustomed to friends talking and thinking apocalyptically. They have friends who work at Anthropic, for instance, “who care deeply about the world going well.” They also know people who see Anthropic as part of the problem.

“I have regularly gone to parties where like half of the party thinks that the other half is driving humanity extinct,” they said. “It makes for a very, very weird social environment.”

This article was produced as part of the RNS/Interfaith America Religion Journalism Fellowship.