An interview with OpenAI Chief Research Officer Mark Chen on the Hugging Face incident, slowing AI development, shifting 5%-10% of compute to safety, and more (Will Douglas Heaven/MIT Technology Review)

OpenAI’s reported emphasis on safety after the Hugging Face containment incident signals that frontier AI development may slow as labs devote a meaningful share of compute to risk reduction and controls. For CIOs, the strategic takeaway is that AI capability gains will increasingly come paired with stronger governance expectations, more scrutiny of model behavior, and potentially longer timelines for enterprise-ready releases. IT organizations should expect tighter vendor assurance requirements, more rigorous testing and monitoring, and a greater need to align AI adoption with risk management and compliance processes.