Elite panic is Solnit's term, developed across her disaster research, for the phenomenon in which institutional authorities—government officials, corporate executives, media—respond to the collapse of normal order with predictions of chaos, looting, and violence that prove almost entirely unfounded. The actual population behavior in disasters is overwhelmingly cooperative. The predicted chaos, when it occurs, is more often produced by the authorities' response to the imagined chaos than by any actual disorder. In New Orleans after Katrina, National Guard troops were deployed to prevent looting that was, in many documented cases, residents sharing supplies from flooded stores. The military response to imagined disorder produced the disorder it was meant to prevent. Elite panic reveals a deep distrust, among those who control institutional power, of the population's capacity for self-organization—a distrust that justifies concentration of authority and preemptive suppression of the spontaneous cooperation that disasters demonstrate is possible.
Elite panic in the AI context takes the form of calls for restrictive governance justified by the claim that the technology is too dangerous for broad deployment—issued, typically, by the same companies and institutions that stand to benefit from regulatory arrangements that raise barriers to entry and concentrate AI capability in existing power centers. The pattern is identical to historical elite panic: the authorities who control infrastructure predict that its democratization will produce chaos, and the prediction justifies maintaining concentrated control. Solnit's framework reveals that the panic is not about the technology's danger but about the threat to existing power arrangements that broad access represents.
The concept operates at multiple scales. At the societal level, elite panic manifests as paternalistic governance—the conviction that ordinary people cannot be trusted to make wise choices about powerful tools and that therefore the tools must be controlled by credentialed experts. At the organizational level, it manifests as resistance to distributed decision-making—the manager who cannot relinquish authority because she believes the team will descend into chaos without her direction. At the classroom level, it manifests as the prohibition of AI tools based on the assumption that students will use them to cheat rather than to learn. In each case, the panic reveals more about the authority's distrust of the population than about the population's actual capacities.
Solnit documents that the communities most capable of self-organization in disasters are often the communities that institutional authorities trust least—poor communities, immigrant communities, communities of color. The competence these communities demonstrate under catastrophic conditions exists under ordinary conditions as well, but it is rendered invisible by institutional arrangements that deny them authority. Elite panic, then, is not merely an error in prediction but a mechanism of power—the systematic denial of capacity to populations whose demonstrated capacity would challenge the legitimacy of concentrated authority.
The term appears throughout A Paradise Built in Hell but is developed most fully in Solnit's accounts of the 1906 San Francisco earthquake (where authorities shot people accused of looting and imposed martial law despite minimal actual disorder) and Hurricane Katrina (where media reports of mass violence proved almost entirely fabricated, while the actual violence was disproportionately perpetrated by authorities against residents). The concept builds on earlier work in disaster sociology—particularly E.L. Quarantelli's research demonstrating that disaster myths (panic, looting, dependency) are systematically contradicted by disaster realities (calm, sharing, self-organization).
Solnit's synthesis adds the political dimension: elite panic is not merely mistaken prediction but motivated reasoning. The authorities who predict chaos benefit from the prediction, because it justifies the concentration of power that the demonstration of popular self-organization would challenge. The panic, then, is performative—it produces the institutional response (military deployment, restrictive regulation, centralized control) that prevents the alternative (mutual aid, distributed governance, cooperative self-organization) from stabilizing.
Predicted Chaos Rarely Materializes. Across disasters, the violence and disorder that authorities predict and media report turn out, upon investigation, to be vastly overstated or entirely fabricated. The actual population behavior is overwhelmingly cooperative, resourceful, and effective.
The Response Produces the Disorder. When chaos does occur, it is often caused by authorities' responses to imagined threats—military interventions disrupting functioning mutual aid networks, property-protection measures preventing life-saving resource sharing, institutional reassertions of control that criminalize the cooperation already underway.
Distrust as Power Mechanism. Elite panic reveals and reproduces the assumption that ordinary people cannot be trusted with power—a self-fulfilling prophecy in which institutional arrangements deny populations the opportunity to demonstrate capacity, then point to the absence of demonstrated capacity as justification for continued denial.
AI Elite Panic. Calls for AI regulation from the companies most positioned to benefit from regulatory capture, paternalistic governance frameworks that assume populations cannot use powerful tools wisely, and restrictions on open-source development justified by imagined catastrophic misuse—all replicate the elite panic pattern, using the rhetoric of safety to justify concentration.
The Alternative Exists. Disaster communities demonstrate that populations are capable of self-organization, collective decision-making, and effective cooperation when institutional constraints are removed. The challenge is not to prevent that self-organization (elite panic's goal) but to support it, creating institutions that sustain cooperation rather than suppress it.