AI companies talk about superintelligent AI as if it is inevitable, but recent safety incidents like OpenAI’s Hugging Face breach illustrate the potential dangers of deploying AI systems that are more capable than humans. So what happens when you can’t reliably control the behavior of these systems?
In this episode of TechCrunch’s Equity podcast, Rebecca Beran is joined by Connor Leahy, an AI researcher, entrepreneur, and current U.S. executive director of the nonprofit organization ControlAI. The group is pushing a far more radical approach to AI safety: blocking companies from developing superintelligence altogether. Leahy explains why he believes the risks have grown too large to be managed through coordination and containment alone, and how what seemed outrageous six months ago is now borne out by a wave of new legislation.
Listen to the full episode to hear more about what’s next.
He explains how he views Frontier AI Labs not just as a profit-seeking company, but rather as a political actor, and what that means for the trillions of dollars that are being poured into building data centers. Why Leahy thinks the real point of no return is when AI builds better AI, potentially creating a runaway cycle of self-improvement The case for international “trust but verify” agreements and why Leahy thinks building super-intelligence capabilities isn’t actually in China’s interests.
Subscribe to Equity on YouTube, Apple Podcasts, Overcast, Spotify, and all casts. You can also follow Equity on X and Threads at @EquityPod.
