The AI industry's opacity is now a competitive asset rather than a liability. World-models companies are hoarding information while sitting on massive capital, treating secrecy as part of their moat, yet the same industry simultaneously demands regulatory clarity on agentic systems in finance and pushes users into data collection through consumer products like Meta's Muse and Vocci's ring. Treasury officials are brokering US-China AI dialogue while Trump's base revolts against his data center push, revealing that AI policy has fractured along lines that don't track traditional political coalitions. Anthropic's IPO valuation remains unsolved not because the math is hard but because nobody knows what these companies actually do, which is precisely the point: the industry wants enough transparency to attract capital and policy favor, but not so much that competitors or regulators can see the actual competitive advantage. The real tell is that conversations about slowing down coexist with TechCrunch selling $200 discounts to Disrupt and startups like ScrollEd pitching textbook disruption to the same conference circuit. When a $249 ring raises privacy questions and an education app gets stage time at a major investor event, but the companies building foundational models won't say what they're training on, you're watching an industry that has learned to manage its narrative while keeping its methods sealed. Australia's government is already planning forty-year economic policy around AI's influence, but the people building it are still deciding what counts as public information.
Sloane Duvall