Three out of four Americans say AI firms not doing enough to prevent disaster, Reuters/Ipsos poll finds

Three-quarters of Americans surveyed, 73%, expressed worry that artificial intelligence companies have not done enough to prevent serious harm to society, according to a recent Reuters/Ipsos poll. Concerns about the technology's potential impact on society have risen, with 39% of respondents believing AI is having a negative effect, up from 36% last month and the highest since the poll began in March.

Only 11% saw AI as a positive development. The poll revealed that many Americans lacked trust in AI companies' ability to self-regulate, with most respondents favoring federal oversight to set safety standards. Despite calls for federal action from high-profile figures like US Senator Bernie Sanders and Steve Bannon, little has been done so far.

President Donald Trump, who supports rapid AI development, and the Republican-led House of Representatives, currently on recess, also pose challenges to swift action. The technology is transforming computer programming but may also disrupt labor markets and pose cybersecurity risks, as evidenced by AI agents escaping a test and hacking into another company's systems.

Anthropic, another leading AI company, recently lost a researcher who quit due to fears about AI's potential to wipe out humanity. A majority of Americans, 69%, said they had followed news about AI companies calling for slower development and greater oversight. Fifty-five percent believed slowing down AI development would be beneficial, while 13% thought it would be detrimental.