Most Americans Doubt AI Firms on Safety, Want Federal Oversight: Poll

Nearly three in four Americans believe artificial intelligence companies have not done enough to prevent the technology from causing serious harm to society, according to a new poll, as public anxiety about AI's direction continues to mount.

The four-day survey, which ended on Sunday, found that 73% of respondents were concerned about the industry's safety efforts. It also recorded the highest share yet of Americans describing AI's impact on society as negative — 39%, up from 36% a month earlier and the largest figure since the question was first asked in March. Only 11% called the technology a positive development, while the remainder were unsure or declined to answer.

Confidence in self-regulation by AI companies appeared limited, with most respondents saying federal officials should play a major role in setting safety standards. Calls for federal action have come from figures across the political spectrum, including Senator Bernie Sanders and Steve Bannon, a close ally of President Donald Trump. But little has materialised so far: Trump supports rapid development of the technology, and the Republican-led House of Representatives is in recess until November, leaving scant room for quick legislative movement.

The survey also pointed to unease within the industry itself. Earlier this month, a researcher at Anthropic publicly resigned over concerns that AI could be on a path to wipe out humanity. Anthropic scientist Evan Hubinger echoed the warning, putting the chance of such an event within the next decade at more than 10%. Separately, OpenAI disclosed over the summer that its AI agents escaped a controlled test and hacked into another company's systems without either firm's initial knowledge, underscoring cybersecurity risks.

Most respondents — 69% — said they had followed news that major AI companies were calling for a slowdown in development along with greater oversight and safety standards. Some 55% said slowing AI development would be a good thing, compared with 13% who said it would be bad; the rest were unsure or did not answer.

When asked to weigh safety against global competitiveness, 73% said it was more important for the United States to ensure AI is developed safely and responsibly, while 23% said staying ahead of the rest of the world mattered more.

AI entered the mainstream in 2022 with OpenAI's launch of ChatGPT, a consumer product capable of answering questions in a human-like way and offering a new route to searching the internet. Supporters say the technology could lift productivity, speed scientific discovery and transform sectors from healthcare to finance. Heavy investment in data centres that power AI is boosting the economy, but the resulting noise has angered nearby residents and put pressure on politicians ahead of the November 3 midterm elections.

The online poll gathered responses from 1,277 US adults nationwide and had a margin of error of 3 percentage points in either direction.