P(doom): Does Silicon Valley think it’s in a death spiral?

On 16^th September 2026, an X account linked to the Department of War posted a photo of US President Donald Trump with a fist raised and overlaid onto the Pentagon captioned, “the United States will NEVER be an effective altruist country.” How did we get here?

Effective altruism et al.: inventing the future

Silicon Valley’s technocrat circles have been radically propelled to new ideologies after the past few years pressed against breakthroughs in Artificial Intelligence. A few voices, including Nick Bostrom, Eliezer Yudkowsky, Peter Thiel and Sam Altman have become mainstays of the philosophy, business, and political world with almost no forewarning.

In 2023, philosopher Émile P. Torres and computer scientist Timnit Gebru coined “TESCREAL” (transhumanism, extropianism, singularitarianism, cosmism, rationalism, effective altruism, longtermism) to label the chronological sequence of philosophies materialising from these spaces, laying out an underlying and overlapping set of subcultures that developed out of the same traditions and assumptions. Moreover, Torres and Gebru warn that these assumptions have been rendered invisible to the public by the nature of the tech industry and its internal circulation, and invisible to TESCREAL’s own proponents by way of complete and unquestioned immersion. The genealogy traces the views taken almost for granted in the present-day among technocrats: transhumanism, extropianism, cosmism, and most worryingly singularitarianism as driving factors for business decisions and government policy alignment.

The Rationalist community seeded a great deal of the movement’s present form with Yudkowsky’s forum LessWrong, which spawned dozens of thought-provoking and hotly-debated thought experiments and utilitarian manifestos, from the concept of infohazards (see Roko’s basilisk) and Pascal’s mugging to there’s no such thing as a tree (phylogenetically) and Yudkowsky’s own Sequences (a blog series spanning cognitive biases, Bayesian statistics etc.). Nick Bostrom, progenitor of ideas such as anthropic bias and existential risk (X-risk), co-authored a landmark paper with Yudowsky titled “The Ethics of Artificial Intelligence.” The pair have become the outspoken basis of how Silicon Valley sees the brain, AI safety, and libertarian interpretations of “maximal good”.

Money where your mouth is

Praise has been showered onto the rationalist community by some of the world’s foremost intellectuals, business magnates, investors and world leaders: Stephen Hawking, Bill Gates, Elon Musk, Peter Thiel, Peter Singer, and – until just last week, apparently – Donald Trump and JD Vance. Effective altruism’s more active spinoff, longtermism, has further led to the creation of various organisations by Will MacAskill and his collaborators. Most notable among these is his venture 80,000 Hours, founded to help high-skilled and socially conscious graduates find high-impact careers to match. The organisation’s work is commendable, but it is also due diligence to note that it declares itself in receipt of over £45 million in donations from private (and sometimes anonymous) donors. One of their most public collaborators was Sam Bankman-Fried of FTX, another EA proponent who was revealed to have committed massive fraud by redirecting close to $8 billion in cryptocurrency to his own trading fund and various ostensibly charitable organisations.

Moreover, there has emerged a powerful countercurrent to the cautious and regulatory approach of EA in the form of e/acc (effective accelerationism), propelled in August and September of 2026 to the forefront of AI safety discussions as companies with trillion-dollar combined market caps elevate this philosophical debate to the stage of international business. Jensen Huang of NVIDIA and Trump have called for rapid AI growth, in lockstep with Attorney General Todd Blanche, who rejects the regulatory role of the Department of Justice. Marc Andreessen, Silicon Valley veteran and Trump advisor, together with Canadian physicist and ex-Google engineer Guillam Verdon, launched the e/acc movement as a direct rejection of “doomer” effective altruists, arguing that long-term growth will follow singularity.

When the magician waves his left hand, look at the Right

Let’s go back to 80,000 Hours and the EA spokespeople (namely, those at Anthropic). Jaan Tallinn, with a stake in both, has spent the past week at the UN General Assembly saying he was “completely terrified” of the direction of the industry. He stands to make potentially billions of US dollars from Anthropic’s expected Initial Public Offering, whenever it comes. He defends the holding: “I’m not investing to profit from LLMs – I’m investing because I’m terrified AI will kill us all.” I am tempted to believe this isn’t untrue. Tallinn is one of hundreds of senior experts in Silicon Valley worried about X-risk, and playing dice with the future – the probability of doom, or P(doom), to put it in odiously Bayesian terms.

Whatever his motives, it is easy to question what game the bulk of the pro-regulation AI camp has been playing on the policy and corporate stage. It is likely that in supporting the creation of a regulatory body, the EA camp will land themselves a seat at the table and accordingly coordinate their own strategic development. Another motivation – as Sam Altman and Trump have both alluded to from opposite camps – is one that concerns National Security. The pro-regulation AI camp seeks to limit Chinese market share, which critics warn could also simultaneously (a) deal a deathblow to open and small-scale models by excluding them from data centre infrastructure and (b) create a critical and highly incorporated network of surviving AI companies that become Too Big to Fail (i.e. having to be being bailed out of unprofitable or predatory pricing strategies in the short term).

No questions asked

Beyond the game of chicken being played by both camps is an underlying layer of mutual understanding – an understanding the general public is neither privy to nor shares in its entirety. All of these massively empowered players (with a few exceptions) are Right-of-Centre Americans whose strategic thinking greatly hinges on the assumption that AI will become powerful enough to control and improve arms systems, management, surveillance, and will dictate scientific progress. Almost all the privileged voices are – interestingly but not surprisingly – physicists, computer scientists, entrepreneurs, venture capitalists, and philosophers of the analytic variety. The most common criticism of all of these groups is their egoism and obsession with efficiency, often based on abstract or unjustified metrics.

Verdon, Bostrom, Elon Musk, and MacAskill’s cofounder Benjamin Todd (as well as other quiet leaders in AI like Jeff Bezos) often loudly mention their training in physics as a signifier of their ability to question from first principles the very foundations of Silicon Valley. One major first principle the movement has never quite been able to shake is an inextricable link to eugenics amplified by the institutional chauvinism of their respective departments. Bostrom infamously claimed on his mailing list that “blacks are stupider than whites” and invoked racial slurs – although he has denounced this past, now 30 years ago. Still, the very foundations of modern transhumanism are derived from social Darwinism; longtermism in particular (as espoused by Musk and MacAskill) always and proudly prioritises abstract future lives and future production over “present-day inconveniences”. The contradiction of this noble thinking is that by-and-large its most visible proponents are bound to deliver quarter-over-quarter returns, another unquestionable first principle in the Silicon Valley paradigm.

The development of these technologies, including the massively promising concepts of brain-machine interfaces and extended healthspans, has been marked by an undercurrent of negligence and poor harm consideration. Musk’s Neuralink tests have had their successes marred by alleged abuses in animal testing; and the most visible ‘healthspan’ engineer Bryan Johnson is a billionaire who has gained notoriety for spending millions on a private team of doctors, drawing his family into biohacking, and having confirmed (if quickly terminated) links to Jeffrey Epstein.

So, does Silicon Valley think it’s in a death spiral? The more pertinent question might be whether it needs the rest of us to believe it is. Sincerity and self-interest aren’t mutually exclusive, but the AI alignment sphere is plagued by a massive conflict of interest among key voices. SNL recently parodied Anthropic CEO Dario Amodei and his brand of ‘cautionary’ development: “AI is not a weapon. It’s a tool... a tool for building weapons. And I urge you to urge me to stop.” With this increasingly visible and contradictory framing, regulation becomes less a question of neutral oversight and more clearly a battle to set the dynamics of entry and existence. Trump’s crusade against effective altruism is yet another manifestation of the public and the market being asked to weigh acceleration against annihilation, with no room to question the consolidation of power in the process. Whether or not an AGI, superintelligence, singularity or strong AI (matching, exceeding, recursively improving beyond, or becoming conscious at the level of human understanding) comes to be, the intellectualised and virulent debate on AI safety currently serves mostly as a smokescreen for down-to-earth conditions and actual, ongoing decay in the fallout of reckless development.