Elon Musk: Humans Will Lose Control of AI Within a Decade
Get the Tech newsletter
Daily tech — startups, AI labs, chips, the launches that shape the next decade. Free.
- Elon Musk predicted AI will surpass the combined intelligence of humanity within about five years, telling The Economist: "There really won't be anything that AI can't do better than humans, apart from being human, perhaps."
- Musk said humans are unlikely to remain in control of AI within the next decade, even as the technology produces "an age of amazing abundance," adding: "I can't see any way to really stop this incredible momentum of AI and robots."
- Musk proposed leading AI developers hold regular safety meetings and grant competing labs limited early access to frontier models so they can identify dangerous capabilities before public release, with governments intervening if risks go unaddressed.
- Musk criticized OpenAI's transformation from a nonprofit into an $800 billion for-profit company, saying it strayed from its original mission — though a California jury rejected his $150 billion lawsuit against OpenAI, Sam Altman, and Greg Brockman in May.
- Musk said Anthropic exists because co-founder Dario Amodei "cares about the future of the world" and no longer trusted OpenAI CEO Sam Altman, calling Amodei "a very principled person" whose team has "not set off my evil detector."
- Musk said he would set aside personal rivalries for safety cooperation: "At the end of the day, if we have to talk, we'll talk."
Why it matters: Musk's call for cross-lab frontier model access is the most concrete AI safety governance proposal yet from a frontier developer, but it comes from someone still building a competing frontier model at xAI — giving OpenAI and Anthropic reason to view his plan with skepticism. His own admission that AI development cannot be slowed means the entire safety framework depends on voluntary cooperation among the labs competing most fiercely, with government intervention only as a backstop if a company fails to act on flagged risks.
