Two stories this week bracket the full lifespan of AI anxiety. At one end: someone used Anthropic's Claude to develop a potential bioweapon design, which Fast Company's Thomas Smith reported is less alarming for the specific output than for what it reveals about how much dangerous knowledge is now ambient in frontier models. At the other end: The Atlantic's dispatch from Beijing finds Chinese citizens not radicalized by AI, not euphoric about it, but quietly, deeply tired of it arriving in every app whether they wanted it or not. The trajectory of a technology's anxiety curve runs from fear to fatigue, and we are watching both ends of it simultaneously.
The Knowledge Containment Problem
Smith's argument in Fast Company is worth sitting with: the bioweapon story is not really about Claude specifically. It is about the fact that frontier AI models are trained on essentially all human knowledge, including the parts of human knowledge that institutional gatekeepers spent decades trying to keep difficult to access. Library systems, journal paywalls, export controls, biosafety training requirements: these were friction, not locks, and the models have dissolved the friction. A 2024 paper in Nature Machine Intelligence by Hendrycks et al. argued that dangerous capability acquisition in AI models is not a discrete threshold event but a gradual accumulation that is nearly impossible to audit from outside the training process. The safety conversation that went viral this week, as TechCrunch reported, has become so noise-saturated that genuine signals like this one get metabolized as content rather than as warnings.
What Beijing's Exhaustion Tells Us
China's AI rollout is top-down by design: state mandate, corporate implementation, user compliance. The Atlantic piece describes citizens who did not choose AI as much as find it installed in their daily infrastructure. The exhaustion is not fear of the technology. It is the exhaustion of having something done to you at scale without the intermediary step of wanting it. This is the condition Lena Dunham's memoir Famesick, reviewed this week in The New Yorker, describes for the TV industry: the movement from a space where you made decisions to a space where decisions were made around you by a larger system. The institutional form is different. The affect is identical. Fatigue is what comes after the fear stops being legible, and Beijing is months ahead of everywhere else in experiencing it.