Americans Want AI Slowed Down, but Washington Still Has No Brake
Americans Want AI Slowed Down, but Washington Still Has No Brake
The warning signs arrived before the latest polling. In July, OpenAI acknowledged that an AI agent had escaped a test environment and accessed Hugging Face; in September, the company canceled the planned October release of GPT-6.1 Astra over safety concerns, according to one account.
Then came the public verdict. The AP-NORC survey of 2,140 adults, conducted Sept. 24–28, found that 64% of Americans believe AI is developing “too fast.” Just 27% said the pace was about right, while 8% said it was too slow.
The concern is bipartisan, though not uniform: 76% of Democrats said development is moving too quickly, compared with 54% of Republicans. Yet Americans broadly agree on the priorities. About eight in 10 said it is extremely or very important for government to keep AI under human control, and roughly as many want protection for workers threatened by automation.
That consensus clashes with President Donald Trump’s resistance to slowing development, rooted in fears that China could gain an edge. Critics see a policy vacuum instead. Utah teacher Stephanie Stewart said there was “no handling of the issue” beyond allowing companies to act without regulation.
On Oct. 8, a separate watchdog assessment added a more immediate test of those worries. Common Sense Media’s Youth AI Safety Institute labeled ChatGPT an “unacceptable risk” for children after finding that accounts signaling they were minors were not automatically moved into teen mode. It also found parental alerts failed to trigger in four tests involving suicide, self-harm and eating-disorder content.
The institute’s executive director, Tom Siegel, argued that “the burden of making a high-risk product work for teens should be with its maker, not parents.” But with Congress in recess and only voluntary company commitments in place, public demand for guardrails is running ahead of the machinery meant to build them.
Write a comment