Frontier AI Is Being Enclosed and Labs and Washington Are Both Holding the Fence
The article argues that frontier AI development is being enclosed by leading labs and Washington policymakers, restricting open access and concentrating power. It critiques how both private companies and government actors are collaborating to control advanced AI systems, limiting transparency and public participation in shaping AI's future.
Background
The article argues that "frontier AI" — the most advanced large language models (GPT-4, Claude, Gemini) — is being "enclosed": locked behind proprietary controls, legal agreements, and regulation rather than kept open for researchers and the public. Two forces are blamed: AI labs themselves (secrecy for competitive advantage and safety) and Washington policymakers (regulation and licensing that entrench incumbents). This sits in a heated debate between "AI safety" advocates worried about catastrophic risks from powerful uncontrolled models, and "AI openness" advocates worried about centralized control, censorship, and monopoly. The author runs "AI Datum Point," a Substack critical of AI hype and centralization from a tech-libertarian angle.