Anthropic Employees Consider Buying Remote Land Amid AI Safety Concerns

In recent weeks, some early employees of Anthropic have discussed plans to purchase remote land in the U.S. as a precautionary measure in case of adverse AI developments. This conversation reflects longstanding concerns in the AI community about uncontrolled AI, a topic first raised by Eliezer Yudkowsky in the mid-2000s. The discussions reportedly include scenarios reminiscent of a Manhattan Project, where employees might be relocated to a desert facility designed to be electromagnetically shielded. The report highlights the connection between Anthropic's founding team and the Effective Altruism subculture, noting that many of these individuals are part of a Bay Area network of 'doomers' who have long contemplated potential catastrophic outcomes related to AI. This trend among Anthropic employees indicates a growing urgency within the AI research community to prepare for worst-case scenarios.