73% of Americans Concerned About AI Existential Risks

A Quinnipiac University national poll released on Wednesday indicates that a significant majority of Americans, specifically 73%, harbor concerns about the potential for future artificial intelligence systems to threaten human survival. This widespread apprehension extends beyond immediate worries like job displacement and increased energy costs associated with new data centers, signaling that fears of AI posing an existential threat have entered mainstream public discourse. The poll found that 38% of respondents are "very concerned" about AI's existential risks, while an additional 35% are "somewhat concerned." Conversely, only 25% of Americans expressed minimal concern, with 14% stating they are "not so concerned" and 11% reporting they are "not concerned at all."

Further dissecting these concerns, the poll revealed that over half of those surveyed, 52%, are more worried about the potential for other humans to misuse AI to inflict harm on humanity. A substantial minority, 33%, expressed greater concern about autonomous AI agents acting independently to cause harm. These findings emerge at a time when both the Trump administration and the AI industry are actively seeking to improve the public perception of artificial intelligence. On Tuesday, President Donald Trump convened a meeting in the Oval Office where executives from leading AI companies signed a "White House Accord on Super Intelligence." This accord, which follows President Trump's directive for the federal government to adopt the term "super intelligence" when referring to AI, signifies a commitment from these companies to implement rigorous internal safety protocols and undergo external audits for their most advanced AI models.

The signatories to the accord include major industry players such as Meta, OpenAI, Anthropic, xAI, Google, and Nvidia. Reports from Semafor suggest that the impetus for this accord originated from a discussion between Meta CEO Mark Zuckerberg and House Speaker Mike Johnson during a state dinner honoring Chinese leader Xi Jinping. Zuckerberg reportedly shared a set of principles with Nvidia CEO Jensen Huang and subsequently circulated a draft of the accord prior to the Tuesday meeting. However, the Quinnipiac poll data suggests that such commitments may not be sufficient to assuage public apprehension. A substantial 74% of respondents indicated a lack of significant trust in the leaders of AI companies, with 29% reporting "not much" trust and 45% stating they have "none at all." Only 21% of those polled expressed "some" trust in AI executives. Furthermore, three-fourths of Americans believe it is "very important" for AI companies to prioritize safety measures, highlighting a disconnect between industry assurances and public expectations regarding AI development and deployment.