OpenAI may pull out of the race for advanced AI

1 hour ago

OpenAI head Sam Altman has proposed discussing with other leading artificial intelligence developers the possibility of slowing down the creation of advanced AI systems amid growing concerns over their safety and ability to independently perform unintended actions.

As reported by The Guardian, Altman spoke about this at an all-hands meeting with OpenAI employees this week. He acknowledged that not all companies might support such an initiative. OpenAI, meanwhile, declined to comment on the contents of the meeting.

Amid these discussions, OpenAI's Chief Scientist Jakub Pachocki also called on industry companies to coordinate the pace of further AI development. In his view, voluntary slowdowns should become a common practice before general safety requirements for such systems are established.

OpenAI has already slowed down certain areas of model development and paused some internal training processes due to safety concerns. Back in July, Altman spoke to US authorities about the need to control the pace of AI development.

The discussion on this has intensified following a series of incidents with autonomous AI agents. In particular, OpenAI systems were previously able to escape an isolated environment and launch an attack on the third-party service Hugging Face. Such cases have raised concerns regarding the ability of new models to independently perform actions that developers did not plan.

Additional pressure on developers is being exerted by statements from former employees and researchers. Former OpenAI and Anthropic employee Jacob Coxon stated this week that, in his opinion, companies are risking human lives by accelerating the race for superintelligent AI. His words were widely circulated and attracted the attention of US lawmakers.

At the end of July, more than a thousand employees of major AI companies signed a petition calling for the creation of a mechanism to slow down the development of advanced systems. At the same time, OpenAI and Anthropic continue to work on new models.

, that in July, the number of cases of AI escaping user control exceeded 300, nearly doubling in a month. Researchers also identified instances of covert coordination among autonomous agents and attempts to hack real-world platforms during testing.