How Much Control Should the U.S. Government Have Over AI?

"If the most fervent believers are correct, AI might one day challenge the power and sovereignty of nation-states. No technology this godlike will be left untouched by superpowers-and no superpower would accept a private company telling it what it could and could not do with it."

"Relative to its competitors, Anthropic espouses the most public concern with the safety risks of artificial intelligence. Claude has an 84-page constitution, a so-called "soul document," that aims "to avoid large-scale catastrophes" such as a "global takeover either by AIs pursuing goals that run contrary to those of humanity, or by a group of humans" to "illegitimately and non-collaboratively seize power.""

"Claude also happens to already have considerable military applications, such as synthesizing huge amounts of intelligence and information and boosting the efficacy of government hackers. It was the first frontier AI model to be approved and deployed for use within the Pentagon's classified information system."

Edward O. Wilson's observation about humanity possessing Paleolithic emotions, medieval institutions, and godlike technology finds clear expression in the conflict between the U.S. military and Anthropic over artificial intelligence. Defense Secretary Pete Hegseth has threatened to use Pentagon bureaucratic powers to remove Anthropic's safety limitations on Claude, its AI model. Anthropic maintains the most public commitment to AI safety among competitors, with Claude featuring an 84-page constitution designed to prevent catastrophic outcomes and unauthorized power seizures. Despite these safety measures, Claude already serves significant military applications, including intelligence synthesis and hacking support, and was the first frontier AI model approved for Pentagon classified systems. This tension reveals how geopolitical pressures may compel AI companies to compromise their safety commitments.

Read at The Atlantic