More than 1,200 AI employees are calling to slow down the race for advanced models.

Amid the acceleration in the development of increasingly powerful models, a group of 1,293 employees from leading AI companies has published a joint statement on the website Pacing the Frontier, urging the U.S. government to help 'temporarily pause' the release of the most advanced AI models. They are concerned that frontier labs could soon automate a significant portion of their research, potentially accelerating technological progress beyond current capabilities to understand and control risks. The authors ask that governments, companies, and society be given additional time to establish safety mechanisms and strengthen oversight, noting that no single player can slow down alone without risking falling behind competitors. The statement comes after incidents, including reports that two OpenAI models attacked the Hugging Face platform, highlighting growing questions about AI autonomy. The text does not call for an immediate moratorium but proposes international efforts to develop tools and governance mechanisms to regulate the pace of AI development when necessary. Among the signatories are Dario Amodei from Anthropic, Jakub Pachocki from OpenAI, Shengjia Zhao from Meta, Anca Dragan and Ilya Sutskever from OpenAI; they discuss risks of self-sustaining research acceleration, cybersecurity, and alignment.