An 'AI freeze' could make big AI companies bigger and hurt smaller firms

Top executives at leading AI labs - including Anthropic's Dario Amodei, OpenAI's Sam Altman, Elon Musk's SpaceXAI, Google DeepMind's Demis Hassabis and Microsoft's Satya Nadella - are publicly calling to slow the pace of advanced model development, arguing that pausing training and releases is needed to address safety risks from autonomous agents and potential loss of human control. Amodei framed the proposal as "pacing the frontier" and urged global coordination and verifiability; supporters differ on mechanisms but share the goal of slowing rapid model progress.

Critics counter that a coordinated slowdown will entrench the largest "frontier" labs and cripple smaller companies and open-source efforts trying to catch up. Startups such as Cohere and Perplexity, researchers and antitrust experts warn a freeze would lock in advantages held by firms with the most compute, capital and distribution, squeeze rivals into acquisitions or failure, and stall competition just as some incumbents eye blockbuster IPOs. A researcher estimated a roughly six-month gap between his institute's model and the leaders', and four consumers have filed a federal antitrust lawsuit alleging an illegal pact among major labs to slow development. Internationally, opponents stress that absent enforceable cooperation with China, a U.S. slowdown would simply let foreign firms surge ahead.