AI safety group pushes for tighter export controls, limits on ‘vastly superhuman AI’

A conservative AI safety group is calling for the U.S. to implement tighter restrictions on advanced chip sales to China and limit the development of “vastly superhuman AI” as part of a new framework for AI policy.

Humans First, chaired by former Tea Party leader Amy Kremer, laid out what it described as an “America First” policy framework.

“AI will supercharge our nation’s economy, improve our quality of life, and ensure we dominate around the world,” the group wrote. “But policy must always be focused on helping hardworking Americans and putting Humans First.”

The group is arguing for stricter export controls on high-end chips and other AI technologies, mandatory testing of models for extreme risks, mandatory public reporting of major AI incidents, developer liability for “reckless design” and a “red line” on self-improving AI.

It also called for an end to any federal use of Flock cameras and warrantless searches of private data, in addition to urging a halt on H-1B visa approvals and opposing federal preemption of state AI laws.

Humans First, which was initially created to organize across party lines, shifted its focus to conservative activists in April when it tapped Kremer to serve as chair.

The group has been a vocal critic of the data center build-out in recent months, leading a day of nationwide demonstrations in July and staging a cross-country bus tour opposing AI infrastructure that wrapped up earlier this month.

The group’s AI framework stands in contrast to the more hands-off approach currently championed by President Trump, who has argued stricter regulations could hamper the U.S.’s ability to compete with China.

Trump gathered major AI leaders at the White House late last month, where they signed on to a series of voluntary commitments to place internal controls on their models and allow for external auditing.