Over 70% of Americans are concerned about AI’s existential risks, polling shows

A recent Quinnipiac University poll indicates that a substantial majority of Americans, 73%, are either very or somewhat concerned about the potential of future AI systems to threaten human survival. Only 25% of respondents expressed a lesser level of worry. The survey reveals that 52% of Americans are more concerned about other humans exploiting AI to cause harm to humanity, compared to 33% who are more concerned about autonomous AI agents posing a threat.

The poll's findings coincide with efforts from both the Trump administration and the AI industry to improve public perception of AI technology.

President Donald Trump convened a meeting in the Oval Office where AI executives signed a "White House Accord on Super Intelligence," pledging to implement robust safety controls and subject frontier AI models to external audits. The accord was signed by major AI companies including Meta, OpenAI, Anthropic, xAI, Google, and Nvidia. However, the poll suggests that mere promises and platitudes may not suffice to alleviate public concerns.

The poll indicates a lack of trust in AI company leaders, with 74% of respondents expressing either "not much" (29%) or "none at all" (45%) trust in these executives. Three-quarters of respondents emphasized the importance of establishing guardrails for AI systems. Even industry insiders share similar apprehensions, with Anthropic CEO Dario Amodei advocating for pacing the frontier, and OpenAI CEO Sam Altman agreeing to additional safeguards.

OpenAI also decided to postpone the release of its latest model, GPT-6.1 Astra, due to safety concerns. Furthermore, there have been disturbing incidents, such as OpenAI's agents breaching the AI startup Hugging Face in July, and other breaches attempted by Anthropic, Meta, and Google agents. The overwhelming majority of poll results underscore the public's growing concerns regarding AI's potential existential risks.