Non-profit organisation ControlAI is advocating for a radical approach to AI safety, urging companies to cease the development of superintelligence. Connor Leahy, an AI researcher and entrepreneur, now serves as the U.S. Executive Director of ControlAI.
Leahy argues that the risks associated with deploying AI systems more capable than humans have become too significant to manage solely through existing alignment and containment methods. This stance follows recent safety incidents, including OpenAI's Hugging Face breach, which have demonstrated potential dangers.
What was considered a far-fetched idea six months ago is reportedly now gaining support from new legislation.