ControlAI, a U.S. nonprofit organization, is advocating for an immediate stop to the development of superintelligent artificial intelligence on the grounds that current safety approaches are inadequate.
Connor Leahy, an AI researcher and entrepreneur who serves as the U.S. Executive Director of ControlAI, discussed the group’s position on the TechCrunch Equity podcast with host Rebecca Bellan. Leahy pointed to recent safety incidents — notably the breach linked to OpenAI and Hugging Face — as examples that illustrate the potential dangers of deploying systems that could surpass human capabilities.
According to Leahy, conventional strategies such as alignment (ensuring AI goals match human values) and containment (technical and procedural limits) are not sufficient by themselves to reliably manage the risks posed by these systems. For that reason, ControlAI is pushing for a more radical approach: preventing companies from continuing development toward superintelligence.
Leahy noted that what sounded far-fetched months ago is gaining traction as high-profile safety failures draw attention. He also said that the proposal is being supported by a wave of new legislation, which has helped increase political and public scrutiny of AI capabilities and safety.
Why this matters
ControlAI’s proposal raises core questions about how to measure and limit the capabilities of AI systems and about the point at which a system becomes too powerful for traditional safety measures to control. The organization argues that the potential harms — underscored by recent incidents — are too great to rely solely on technical fixes.
The recommendation to halt development is likely to prompt debate among researchers, companies, and lawmakers, because pausing progress toward superintelligence would carry significant economic and scientific consequences. At the same time, ControlAI’s emergence highlights accelerating societal and regulatory attention to AI safety.
On TechCrunch’s Equity podcast, the conversation with Rebecca Bellan explored how ControlAI’s stance fits into ongoing discussions about safety and governance, and how public discourse is shifting in response to further incidents and legislative initiatives.
Next steps
Leahy’s and ControlAI’s proposals are expected to fuel further debate and investigation into AI development pathways, the need for regulation, and what mechanisms might be necessary to ensure the technology does not create uncontrollable risks.
