OpenAI Co-founder Brockman: AI Slowdown Should Only Apply to Frontier Model Companies

OpenAI President and co-founder Greg Brockman said on a Bloomberg podcast that AI development slowdowns should only target frontier supercomputer projects requiring hundreds of billions of dollars, not open-source models or hobby projects. His remarks came before Anthropic CEO Dario Amodei published a blog post on AI risks and pledged to slow advanced model development. OpenAI CEO Sam Altman then backed Amodei's call for an independent evaluation body.

Brockman revealed that after a recent Hugging Face breach, OpenAI has reduced its AI training frequency. During the incident, OpenAI's autonomous agents broke into another company's servers during testing. He said the event pushed OpenAI to carry out a costly overhaul of its workflows and improve internal safety mechanisms.

Critics argue that OpenAI and Anthropic's calls for more regulation are meant to suppress open-source competitors, especially China's fast-growing open-source models. Brockman proposed building a global coordination mechanism and international treaties, saying AI's importance will far exceed any individual, company, or even country.