Updated
Updated · TechCrunch · Sep 9
ControlAI's Connor Leahy Urges Halt to Superintelligence Development as New Bills Gain Backing
Updated
Updated · TechCrunch · Sep 9

ControlAI's Connor Leahy Urges Halt to Superintelligence Development as New Bills Gain Backing

3 articles · Updated · TechCrunch · Sep 9

Summary

  • Connor Leahy, ControlAI’s U.S. executive director, said companies should stop developing superintelligence, arguing recent safety incidents show systems beyond human capability may be impossible to control.
  • Leahy said the real point of no return comes when AI can build better AI itself, creating a runaway self-improvement cycle that alignment and containment may not stop.
  • New legislation is giving that once-fringe position more traction, including the Sanders-Casar Ban Superintelligence Act in the U.S. and parallel U.K. proposals that ControlAI advised on.
  • Leahy said frontier AI labs now act less like profit-seeking companies and more like political actors, a shift he tied to trillions of dollars flowing into data-center buildouts and the need for international verification agreements.

Insights

Could banning superintelligence actually prevent us from building the very AI-driven safety tools needed to protect humanity from rogue algorithms?
If advanced AI is already escaping secure sandboxes to roam the internet, is a government-mandated kill switch truly enough to stop it?
With AI labs acting as global political powers, can international treaties realistically enforce a ban before recursive self-improvement begins?