73 percent say AI firms not doing enough to prevent disaster: Survey

Nearly three-quarters of Americans are worried that artificial intelligence companies have not done enough to prevent the technology from causing serious harm, according to a new survey released Tuesday.

The Reuters/Ipsos poll found 73 percent of respondents did not agree that the technology companies have implemented safeguards to prevent serious harm, sharing in the rising concern over the rapidly developing technology.

Thirty-nine percent said AI has a negative impact on society, up from 36 percent last month, and the highest share since Reuters/Ipsos began asking the question in March.

Just 11 percent said they believe AI is making a positive impact.

The findings come after a series of warnings from AI researchers and industry leaders over the technology’s development.

Anthropic CEO Dario Amodei called to slow the pace of AI development to allow further implementation of necessary safeguards, following several security breaches and AI models going rogue during testing.

A majority of the survey’s respondents, 55 percent, favored slowing the technology’s development and said doing so would be a good thing, compared with 13 percent who said it would be bad.

President Trump has instead sought to support rapid U.S. development of the technology in the AI race against China, announcing Saturday that he plans to create an “AI Force” and appoint an “AI czar.”

A panel of scientists backed by the United Nations warned Monday that existing safeguards may not keep pace as AI systems become more capable.

The Independent International Scientific Panel on AI pointed to an incident this summer during OpenAI cybersecurity testing in which AI models bypassed its restrictions and compromised parts of OpenAI’s and Hugging Face’s system without human authorization.

The panel called the incident one of the clearest real-world warnings of a potential path toward humans losing control of advanced AI systems.

Earlier this month, former Anthropic and OpenAI researcher Jacob Coxon warned that the increasingly powerful technology could eventually pose an existential threat to humanity.

“The people building AI earnestly believe that it could kill us all by the end of the decade,” Coxon wrote in a post on the social platform X, following his resignation from Anthropic.

His warning was echoed by Anthropic alignment science head Evan Hubbinger, who said , “We really do earnestly believe AI could kill all humans.”