AI companies have been talking about superintelligent AI as if it were inevitable, but recent security incidents like OpenAI Hugs Breach of face are demonstrating the potential dangers of implementing AI systems that are more capable than humans. So what happens when we can’t reliably control what these systems do?
In this episode of TechCrunch Equity podcast, Rebecca Bellan is joined by Connor LeahyAI researcher, entrepreneur and now CEO of a non-profit organization in the US. ControlAIwhich is driving a much more radical approach to AI security: completely preventing companies from developing superintelligence. Leahy explains why he believes the risks have become too great to manage through alignment and containment alone, and how what seemed far-fetched six months ago is now backed by a wave of new legislation.
Subscribe to Equity at YouTube, Apple Podcasts, Cloudy, Spotify and all the casts. You can also follow Equity on unknown and Ragsat @EquityPod.