Poll Finds 63% of Americans See at Least Moderate Risk From Advanced AI

A new survey indicates that concerns about artificial intelligence extend across political groups, with 63% of Americans saying there is at least a moderate risk that advanced AI could destroy humanity, Andrew Romano reported for Yahoo Tech.

A new U.S. poll found widespread concern about the potential risks of increasingly advanced artificial intelligence systems.

The Politico poll found that 63% of respondents viewed the risk as at least moderate.

That included 70% of respondents who voted for former Vice President Kamala Harris in 2024 and 60% of those who voted for President Donald Trump, according to the survey.

Seven percent said there was no risk at all. Thirty-seven percent described the risk as either significant or almost certain.

The survey, conducted Sept. 13-15, also found that 48% of U.S. adults agreed that AI development had reached a point at which development of more advanced models should be paused to reduce potential risks.

Thirty-one percent favored continuing to develop more advanced AI because of the potential benefits, while the remainder did not choose either position.

The survey also found that 64% of Americans considered it somewhat or very likely that AI would eventually be able to perform any intellectual task a human can, while 63% said it was somewhat or very likely that AI could become so advanced that humans would no longer be able to control it.

The findings come amid an intensifying debate over AI safety.

Anthropic and OpenAI executives have called for additional safeguards and measures to manage the development of frontier AI, while other technology leaders have opposed broad restrictions or a coordinated slowdown.