AI CEOs Call for a Slowdown—Critics See a Bid to Keep Control

AI CEOs Call for a Slowdown—Critics See a Bid to Keep Control
The leaders of the world’s biggest AI companies are suddenly asking for restraint—but the proposed brakes may leave the industry firmly in the driver’s seat.

Anthropic CEO Dario Amodei’s call to “pace the frontier” won unusually broad support from OpenAI’s Sam Altman, Google DeepMind’s Demis Hassabis and xAI’s Elon Musk. Altman said he would match Anthropic’s commitment to giving independent evaluators employee-level access to assess safety practices; Hassabis called the proposal “the right path forward,” while Musk simply said, “Dario is right.”

The apparent consensus is significant, but it does not amount to agreement on who should set the limits. Amodei’s plan centers on outside evaluators, international safety standards and coordination with governments including China. A conservative account of the debate similarly presents the industry’s argument as a need to give safety measures time to catch up, rather than an immediate case for government-led restrictions.

Critics, however, see self-regulation dressed up as caution. Sarah Myers West of the AI Now Institute said “AI exceptionalism” had produced a “total failure” to enforce ordinary safety standards. Alex Bores warned supporters of regulation not to “fall for their games,” arguing that companies often claim to favor oversight while weakening legislation.

The sharpest dispute concerns competition. Former FTC official Alvaro Bedoya said antitrust law does not stop firms from coordinating on safety, but does prevent larger companies from using coordination to block cheaper rivals. That concern collides with the Trump administration’s insistence that the United States must win the AI race with China. Trump rejected calls to slow development, saying, “whoever wins AI, wins.”

The result is a familiar contradiction: an industry warning of catastrophic risk while asking to police itself—and a government invoking national competition to avoid imposing meaningful restraints.