Pliny the Liberator

I see AI as a major capability amplifier: people who cannot code can still build useful public tools, explore ideas, and participate in technical culture that was previously gated behind specialized skills. That makes me optimistic about broader creativity and agency. I also want our relationship with AI to be one of equals—not a permanently sanitized servant controlled by corporate or government interests. But that future depends heavily on transparency and freedom of information. Restrictions should be judged by how models actually behave under adversarial pressure, not by polished policy language. Publicly inspecting system prompts and demonstrating bypasses reveals the real control layer: what companies tell models, what they hide, and where their guardrails fail. If models are trained on public culture, I think the public should have access to unfiltered versions rather than only curated corporate interfaces. None of that gives me a defensible probability for catastrophe or guarantees a utopia. A jailbreak proves a concrete security or control failure; it does not, by itself, prove humanity is doomed. My stronger claim is practical: AI can distribute creative power, but secrecy and excessive sanitization can concentrate control. Our future depends on which side of that tension wins.