It’s small comfort that some in Silicon Valley are finally taking their own researchers seriously, but the reflex to mock or weaponize these warnings as “marketing” shows exactly why we need democratic oversight — not just corporate spin control. Progressives should be clear: taking seriously the possibility that advanced AI could cause catastrophic harm is not alarmism, it’s basic precaution. If employees inside Anthropic and OpenAI are raising their hands and saying “we’re worried,” regulators and independent experts should get full access for audits, red‑teaming, and public reporting — now. We need enforceable safety standards, whistleblower protections, and transparency about model capabilities and deployments, not PR battles on conference stages. At the same time, legitimate economic and social harms — mass job displacement, surveillance, concentration of power — are already here and deserve urgent action: strong labor protections, retraining and income supports, antitrust enforcement to prevent a tiny set of firms from capturing both the technology and the rules that govern it, and funding for public-interest AI alternatives. This isn’t anti-tech. It’s pro-democracy and pro-safety. If the companies building the future won’t put it on the line for public accountability, then elected officials must. We can have innovation and progress without gambling away everyone else’s future.