OpenAI CEO Altman Warns Unchecked AI Could Widen Inequality, Urges Speed Limits on Recursive Self-Improvement

OpenAI CEO Sam Altman warned that unconstrained AI development could deepen social inequality and said global limits are needed on the pace of recursive self-improvement (RSI), the process by which AI systems improve their own capabilities, in an interview with tech YouTuber Matthew Berman published by the Forward Future podcast. Altman said a lab pushing RSI at full speed could achieve the capability but would carry major risks, arguing for a mechanism to cap how fast it advances. The warning underscores a tension inside OpenAI, which continues to fund models, chips and data centers while Altman insists technology leadership cannot be the company's only advantage. He cited customer relationships, network effects and cost-efficient infrastructure as durable moats. Altman said current models are not conscious but rejected granting AI legal personhood or subjective will, stressing humans must retain decision-making authority. He also backed reform of FDA approval processes while affirming existing safety testing. On compute, Altman said demand has no clear ceiling, citing a Jevons Paradox effect in which cheaper intelligence drives more uses. OpenAI is developing in-house chips and new consumer hardware.