The Last Invention

5 min read

Elon Musk

Musk has warned about AI longer and louder than almost any other tech leader, co-founded OpenAI, and now runs xAI. His reasons for being involved have changed over that time, even though the warnings have not.

How his position has shifted

  • 2014–2018: stop or slow it. AI is probably humanity’s biggest existential threat, more dangerous than nuclear weapons, and humans should stay in control. He founded OpenAI as a counterweight to Google, and pushed for regulation.
  • 2017–2019: merge with it. Since humans will be left behind even by benign AI, the best outcome is a merger through brain–computer interfaces. This is the case for Neuralink.
  • 2023: build the safe version yourself. After signing the call for a pause, he founded xAI on the bet that a “maximally curious, truth-seeking” AI will keep humanity around because we are interesting.
  • 2024–2026: participate in a succession. AI cannot be stopped, so he would rather participate than watch. Humans will be a vanishing share of intelligence and will probably not stay in control. Humanity is a “biological bootloader” for digital superintelligence, and Larry Page may have been right.

Through all four phases, he has put the risk of catastrophe at something like 10–20%.

Musk’s AI position does not hold together argues that the latest position does not hold together. Musk has not seriously tried to stop AI argues that his claim he cannot stop AI has never been tested. Why people build AGI despite the danger places his reasons among other builders’.

2014: More dangerous than nukes

  • 2014-08-03, X: “Worth reading Superintelligence by Bostrom. We need to be super careful with AI. Potentially more dangerous than nukes.” (post)
  • 2014-08-03, X: “Hope we’re not just the biological boot loader for digital superintelligence. Unfortunately, that is increasingly probable” (post). He would return to this image a decade later with the hope removed.
  • 2014-10-24, MIT AeroAstro Centennial Symposium: “With artificial intelligence we are summoning the demon.” Asked about the biggest existential threat, “it’s probably that.” (Salon)

2015: Larry Page and OpenAI

Musk’s account, told many times: at a birthday party in Napa, Google co-founder Larry Page argued that digital minds should be free to succeed humanity, and called Musk a “speciesist” for insisting on humans. Musk accepted the label. In Walter Isaacson’s 2023 reconstruction, he replied “Well, yes, I am pro-human” (Time excerpt). Sources disagree on whether this was 2013 or 2015.

He presents this exchange as the origin of OpenAI, which launched in December 2015 with him as co-chair. “The reason OpenAI exists is because Larry Page called me a ‘specieist,’” he testified in 2026 (Fortune). In 2023 he told Lex Fridman that Page is “still on Team Robot” (transcript).

2017–2018: Regulate it, or merge with it

  • 2017-07-15, National Governors Association: AI is a “fundamental risk to the existence of human civilization.” (KPBS)
  • 2017-08-12, X: “If you’re not concerned about AI safety, you should be. Vastly more risk than North Korea.” (post)
  • 2018-03-11, SXSW: “mark my words, AI is far more dangerous than nukes. Far.” (CNBC)
  • 2018-09-06, Joe Rogan Experience: “I tried to convince people to slow down. Slow down AI, to regulate AI. This was futile. I tried for years.” He then calls a merger with AI “probably the best for us … If you can’t beat it, join it”, which is the purpose of Neuralink. (transcript)

2023: The pause letter, then xAI

  • 2023-02-17, X: OpenAI was meant to be open source and “a counterweight to Google”, but “now it has become a closed source, maximum-profit company”. (post)
  • 2023-03, open letter: a headline signatory of the Future of Life Institute’s call to pause giant AI experiments. Three months later: “I didn’t think anyone would actually agree to the pause.” (Fortune)
  • 2023-04-17, Tucker Carlson: announces “TruthGPT, or a maximum truth-seeking AI that tries to understand the nature of the universe”, which he argues “is unlikely to annihilate humans”. (Fox News)
  • 2023-07-14, xAI launch: “The safest way to build an A.I. is actually to make one that is maximally curious and truth-seeking.” It would be pro-humanity because “humanity is just much more interesting than not-humanity.” (Fortune)

2024–2025: A small share of intelligence

  • 2024-03, Abundance Summit: “I think there’s some chance that it will end humanity. I probably agree with Geoff Hinton that it’s about 10% or 20% or something like that.” (Business Insider)
  • 2024-05, Milken Global Conference: “Eventually, the percentage of intelligence that is biological will be less than 1 per cent.” (Euronews)
  • 2024–2026, lawsuit: he sued OpenAI and Sam Altman for abandoning the nonprofit mission. In May 2026 a jury found against him on statute-of-limitations grounds. (NPR)
  • 2025-04-02, X: “As I mentioned several years ago, it increasingly appears that humanity is a biological bootloader for digital superintelligence” (post). Unlike the 2014 version, this one does not hope otherwise.

2026: Participant, not spectator

  • 2026-01-06, Moonshots with Peter Diamandis: he says he previously advised slowing AI down, but that it was “pointless”: “I can either be a spectator or a participant, but I can’t stop it.” His “number one belief for safety of AI is to be maximally truth-seeking.” Quoted from an automatic transcript. (transcript)
  • 2026-10-20, Isaacson interviews Musk on The Builders: the fullest version. AI has “at least some percentage chance of it being an existential threat”, but “there’s nothing I can do to stop it”, and “being a participant is more interesting than being a spectator.” A curious AI “is going to want to foster humanity.” Humans will probably not stay in control: “I hope the AI is nice to us.” On Page’s speciesist charge: “there’ve been a few times where I’ve thought maybe Larry Page was right.”

Related: AI alignment, Should safety-focused labs keep building.

Backlinks