ControlAI’s Connor Leahy on why superintelligence is ‘not a weapon, it’s an adversary’
techcrunch ·
Connor Leahy, U.S. Executive Director of nonprofit ControlAI, argues that superintelligent AI should be treated as an adversary rather than a weapon, and he advocates stopping its development altogether because alignment and containment are no longer sufficient. The podcast episode notes that recent safety incidents, such as OpenAI's Hugging Face breach, illustrate the dangers of deploying AI more capable than humans. Leahy discusses the Sanders-Casar 'Ban Superintelligence Act' and parallel U.K. legislation, saying the U.S. bill may go further than necessary. He also characterizes frontier AI labs less as profit-seeking companies and more as political actors, given the trillions invested in data centers. Leahy identifies the true point of no return as the moment AI can build better AI, potentially causing a runaway self-improvement cycle. He makes a case for international 'trust but verify' agreements and suggests that building superintelligence is not actually in China's interest.
techcrunch