Sam Altman’s evolving stance on AI licensing raises questions about competition and convenience

Three years ago, Sam Altman sat in front of US senators and made a request that surprised much of Silicon Valley: please regulate us. Specifically, he proposed a government-issued licensing regime for AI models that exceed certain capability thresholds, something modeled after nuclear oversight where an authority grants and revokes approvals. Fast forward to June 2026, and Altman is singing a very different tune. He’s now actively lobbying against mandatory pre-release approvals for AI models, instead pushing for increased federal funding directed at testing and evaluation capabilities, particularly at the Department of Commerce. From nuclear-grade oversight to voluntary check-ins On June 2, 2026, President Donald Trump signed an executive order establishing a voluntary framework for AI model testing. The order explicitly rejected mandatory licensing, drawing a clear line against the kind of regime Altman himself had once championed. Rather than push back on the administration’s lighter touch, Altman leaned into it. His new preferred approach centers on giving federal agencies more resources to evaluate AI systems, without granting them the power to block a model’s release before i...