OpenAI's Chief Scientist Calls for Industry to Restrict Model Development; Altman Expresses Agreement

date
23/11/2026
OpenAI chief scientist Jakub Pachocki proposed that AI companies should agree to set limits on developing increasingly capable models until researchers have better ways to understand and monitor the behavior of these models. Pachocki called for industry-wide standards to set boundaries for AI development, with companies agreeing not to cross these boundaries until specific safety requirements are met. "I further propose setting common limits based on perplexity and other metrics, with developers committing not to cross these limits until our ability to understand and monitor AI behavior reaches a specific milestone," Pachocki wrote. The limits proposed by Pachocki could be based in part on a technical metric known as perplexity, which measures an AI model's ability to predict the next segment of text. Pachocki did not specify what level the initial limits should be set at, nor when they should begin to be implemented. Pachocki warned that as early as 2027, AI could accelerate its own evolution at a pace that may outstrip researchers' ability to improve monitoring and understanding of the technology. OpenAI CEO Altman expressed support for the proposal, calling Pachocki's article "the most important thing OpenAI has published in a while." In a post on X, he said that agreeing on common limits would help ensure "safety and control."