Artificial Intelligence
Anthropic veterans consider buying remote land over AI risk
A Wall Street Journal report said some of Anthropic's earliest employees were discussing land in remote parts of the United States as a fallback, reflecting long-running fears in Effective Altruism circles about uncontrolled AI.
- The Wall Street Journal reported that some of Anthropic's earliest employees had been considering buying land in isolated parts of the United States.
- The report said those employees were thinking about where they could move if AI turned dangerous.
- Former employees said similar ideas were discussed at Anthropic dinners in the company's early period.
- The article said early hires at Anthropic and OpenAI were closely linked to Effective Altruism and to a Bay Area circle focused on catastrophic risks.
- It said Eliezer Yudkowsky had warned about uncontrolled AI in the mid-2000s.
The Wall Street Journal said some of Anthropic's earliest employees had been talking over the past few weeks about buying land in remote parts of the United States. The report said this was a fallback in case AI became uncontrollable.
Former employees said the idea of escaping AI was not new inside the company. They recalled early dinners where people discussed being moved to the desert and continuing work inside a shielded government site.
The article linked Anthropic's and OpenAI's early hires to Effective Altruism. It also described a Bay Area network of people who had spent years thinking through catastrophic outcomes.
The piece said Eliezer Yudkowsky had been warning about uncontrolled AI since the mid-2000s. It added that, before AI became the main concern, the same group had focused on dangers such as asteroid impacts and supervolcanoes.
Named in this story
People
- Eliezer Yudkowsky
- warned about uncontrolled AI in the mid-2000s
Companies
- Anthropic
- had some of its earliest employees discussing contingency plans
- OpenAI
- was named alongside Anthropic as having early hires linked to the same subculture
Organisations
- Effective Altruism
- was described as the subculture connecting some early hires
Places
- Bay Area
- was the base of the network of doomer thinkers
- United States
- was where remote land was reportedly being considered
How the source tells it
The register was anxious and speculative, using catastrophe imagery and future-fear scenarios to frame the reporting.
- alarm the piece cast AI as a possible trigger for urgent contingency planning
- doomsday framing it relied on end-of-the-world scenarios and historical disaster analogies
- speculation presented as fact it turned recalled conversations and hypothetical relocation plans into a concrete narrative