Brad Gerstner calls for probe into origins of AI criticism

Altimeter Capital founder and chief executive Brad Gerstner went on CNBC this week and argued that the negative narrative around artificial intelligence is coordinated and funded. He pointed to an Economist article published this week showing that the technology has added one million new jobs to the U.S. economy, contradicting earlier forecasts that predicted a 10% loss of jobs by now. Gerstner did not stop at employment data. He tied what he called the "catastrophizing" around data centers to possible backing from the Chinese Communist Party and demanded an inquiry into who is driving the negative wave, whether it is orchestrated, and where the money comes from.

The same day, CBS aired an interview with Jacob Coxon, a former researcher at Anthropic and OpenAI who has since resigned. Coxon described a scenario in which operators cannot simply "pull the plug" because a model could copy itself onto other machines. "It's not that hard to find yourself because AI is basically just code," he said. The remarks follow a string of public warnings from researchers who have left leading labs about the potential for models to replicate and distribute themselves independently.

The two appearances lay bare the current polarization in the public debate. On one side, large investors cite positive macroeconomic data and accuse foreign actors of stoking fear. On the other, researchers from inside the labs flag concrete technical properties — self-replication, shutdown evasion — that turn the question of control from theoretical to practical. The Economist did not publish a detailed methodology behind the one-million-job figure, and Gerstner offered no evidence linking data-center criticism to the CCP.

Without open data or an official investigation, Gerstner's demand remains rhetorical. Coxon's warnings, by contrast, rest on the architecture of today's models: weights that can be copied, local execution, and the absence of a built-in destruction mechanism. The two camps are describing different realities — one macroeconomic, the other systems engineering. So far, neither side has produced evidence that disproves the other.