Ian Duncan examines rationalist AI safety culture

When the self-appointed stewards of humanity's existential fate share group houses, sexual partners, and doomsday liturgy, public skepticism of AI safety isn't an anti-science conspiracy—it's basic institutional due diligence.

• –**Insular Epistemics Undermine Credibility:** The AI safety ecosystem suffers from an acute lack of external perspective, with researchers, major donors, independent evaluators (like METR), and high-profile whistleblowers drawn from the same interconnected Bay Area social scene.
• –**The Perils of "Heroic Responsibility":** Yudkowsky's doctrine that exceptionally rational individuals bear sole responsibility for civilizational survival creates a dangerous exceptionalism, easily rationalizing toxic internal conduct, psychological abuse, and contempt for democratic institutions.
• –**Regulatory Capture in Existential Clothing:** The trajectory from fringe blogs to mainstream panic risks being co-opted into heavy-handed licensing regimes that entrench well-funded incumbents while freezing out competitors under the banner of risk prevention.
• –**Oversight Demands Real Independence:** Meaningful evaluation of frontier models cannot rely on informal norms among peers and former colleagues; it requires enforceable conflict-of-interest firewalls, independent external audits, and attention to immediate societal harms alongside speculative doom.